Fun little project his weekend, building a object detection and classification solution for less than $100. Though this pic only shows “person” and “book” classifications, the model can classify some 90 objects! The Tensorflow Lite model is running on a 4GB Raspberry Pi 4 w/ 128GB Sdcard. The camera is a Arducam, which I need to work on the resolution for but it didn’t impact the detection or classification, and ran at ~2.0 fps. Running on a Pi I have a give and take between model performance and accuracy, given the limited resources, but will push to see how resource hungry a model I can run on it. More to come…
Wondering what’s real about artificial intelligence? Today on BrainTrust LIVE, we’re fortunate to have Cynthia Holcomb, founder/CEO of Prefeye, and Shawn Harris, Customer Partnerships & Strategy, SmartLens — two retail practitioners who are working with their clients on real A.I. solutions. They’ll give us the lowdown — more specifically, on how retailers can currently use AI for personalization, the limitations that are frustrating them at present, and what does the future holds.
Recording on Facebook: https://www.facebook.com/retailwire/videos/745383885894757/
I enjoyed doing this segment with Jeff Lahens, my former business partner, and Jarvis Green, good friend and 2X Super Bowl champion with The New England Patriots… now owner of Oceans 97.
Module 6 of 6 of MIT CSAIL AI Implications & Strategy: The future of artificial intelligence – Notes
Module 6: The future of artificial intelligence
The future progress of AI
- Promising near-term applications of AI
- machine learning will drive field for next few years, its a new kind of statistics
- Easy to predict the future, hard to say when it will happen
- commercial value will focus on making new things possible, that were not possible before. It’s not just about eliminating jobs
- AI technologies supporting autonomous driving.
- “blunder” stopping will be a huge benefit, in general.
- diagnosis, monitoring and treatment personalize
- move people and goods in more efficient ways
- Keep information private and safe
- Expansions in education.
- Promising areas of progress in AI
- AI helping cognitive tasks, and robotics helping our physical tasks.
- Machines are not good at figuring things out, without labeled data. We will see more ML systems that are able to learn more small data or in the moment interactions in the world.
- Machines need to advance in common sense knowledge
- Human/machine interactions
- make new machines faster
- Interpretabilty is important the ability to explain what they are doing, what’s happening to them, how are they reaching their conclusions
- What will be especially hard for AI
- Interaction with physical world. easier to send a robot to mars, than to clear your dinner table.
- Creating more capable robot bodies and controllers
- We do routine things, they we learn to do routine things… the latter take common sense.
- Adoption rates for technology
- Things take a long time… FYI – “No hands across America” first robotics vehicle drove coast to coast in 1995. oh, and Amazon launched in 1995…it’s only now that they are gaining dominance. However, adoption rates for new tech are plummeting…
- Concerns surrounding human-level AI
- Marvin Minsky “Keep us as pets” ~1970 Life Magazine
- Current systems appear to be human like way, but they do not know what they are doing. Most systems are carefully crafted, AlphaGo cant play chess…
- All current day solutions are point solutions.
- People feel like the AI apocalypse is upon us vs Ai will solve all of the worlds problem we are not there…
- How far away is general AI?
- Excitement has brought a lot of people in to the field.
- One day we will have computers that will have general intelligence.
- it’s far away…
- Andrew Ng… “Worrying about machines taking over is like worrying about overcrowding on Mars.”
- Amara’s Law: “We tend to overestimate the effect of a technology in the short run and underestimate the effect in the long run.” ~Roy Amara
- The Seven Deadly Sins of AI Predictions
- Rodney Brooks is a former director of the Computer Science and Artificial Intelligence Laboratory at MIT and a founder of Rethink Robotics and iRobot.
- We are able to discover things that never could have been before.
- we are figuring out our intelligence… wow.
- Business, enabling discovery in all fields and powering the economy.
- People today are developing the science, engineering, and application of intelligence. wow.
- Clarke’s Three Laws:
- “When a distinguished but elderly scientist states that something is possible, he is almost certainly right. When he states that something is impossible, he is very probably wrong.” ~Arthur C. Clarke
- “The only way of discovering the limits of the possible is to venture a little way past them into the impossible.” ~Arthur C. Clarke
- “Any sufficiently advanced technology is indistinguishable from magic.” ~Arthur C. Clarke
What makes a group smart?
- Measuring the collective intelligence of groups
- Used IQ test to measure collective intelligence of groups, the collective intelligence factors, only moderately correlated to the average intelligence in the groups
- Factors correlating with collective intelligence
- Average social perceptiveness, or social intelligence… reading the minds and the eyes,, guessing emotions.
- Equal Participation in group
- More women, was correlated with more intelligent groups. tied to social perceptiveness
- high social perceptiveness
- Face-to-face interaction vs typing text
- In both cases social perceptiveness lead to greater collective intelligence.
- social intelligence, helps you to better work with other people.
- Two requirements for smarter groups
- Need smart individuals, with abilities to accomplish basic tasks
- Need to be able to work well together
- People: interpersonal skills
- Computers: design computer that can work well with people.
Module 5 of 6 of MIT CSAIL AI Implications & Strategy: Artificial intelligence in business and society – Notes
Module 5: Artificial intelligence in business and society
Professor Patrick Winston gives a detailed example of another type of AI, discussing some of his recent research on building AI systems that learn in much the same way that humans often do: by understanding stories.
- Rule based systems,
- searching many possibilities
- linear regression, probabilistic reasoning
- 5 to 10 years… What’s possible with data and deep nets
- algorithms are free… data is the asset
- AI Forth Wave
- Massive, free computing
- Excited people
- Emerging round table
- Accumulated progress
- Better questions
- What makes us humans different than other specieis, past and present?
- Human have been around 200k years
- about 70k began to advance
- We can take two disparate concepts, and make a new concept without impacting the original two concepts.
- “Merge operation” give us an inner-language…. only we can tell stories
- Artificial Perception
- Story understanding
- Recipe following:
- Strong Story Hypotheses:
- self awareness is a suitcase term
- `System will be able to explain what they do and why?
- machines and programs will tell their own stories.
- smarter application
- applications that can explain themselves
- applications that understand us
- better understanding of ourselves and each other, and that will take us to a new level.
In Module 4, you learned about robotics, which was defined as the automation of physical tasks. However, some people are also using the term “robot” to refer to systems that automate certain kinds of purely information-processing tasks. For instance, the financial world refers to so-called “robo-advisers” that help clients manage their investments, sometimes with a human adviser in the loop and sometimes not.
MIT Professor Andrew Lo discusses some of his research on AI and investment management. He uses the example of index funds to illustrate how algorithms can be applied to the financial world. He introduces the idea of “precision indexes,” which (akin to “personalized medicine” that is specific to a given individual) are automated portfolios that take an individual’s personalized criteria into account to make decisions. Professor Lo notes that organizations are currently missing the opportunity to model actual human behavior, rather than modeling assumptions about what human behavior should be in the investment world. He closes by elaborating on the notion of “bounded rationality.”
- index funds… assets held in proportion to their market capitalization
- People tried to do equal balances, but it was difficult to manage
- Three new criteria for index funds
- If you look at a spectrum from index to hedge, what if you could take a full spectrum approach…”Precision Indexes” E.g. the Shawn Harris 500, based on my particulars… The hardware and software exists today to do this, what we don’t have yet is the algorithms… See “Personal indexes” paper.
- We are missing the ability to model actual behavior, not just modeled behaviors… artificial stupidity…artificial humanity… learned common sense…
- Bounded Rationality
- We don’t know what the optimal solution is, we develop rules of thumbs that are good enough.
- 4 Themes:
- Evolution models of behavior
- surveyor of investor risk preference
- heuristics and algorithms to automate systems
- learn from big data to learn from actual behaviors
The Future of Work
AI and robots are set to play a big role in the future workforce as collaboration between people and computers increases. Although the media seems to relish reporting that robots will replace human workers, it is more likely that people’s jobs will change and evolve, so that people work alongside AI and focus their energies on the tasks they do best.
O-ring principle as a collection of tasks that need to be done together to successfully accomplish a main task. If some of the tasks involved can be automated, the economic value of the human inputs for the other tasks that can’t be done by machines will increase. O-ring Harvard Economist Michael Cramer… As you improve the reliability of other components, other items become more important…. Humans becoming the O-rings.
Never Enough Principle “Insatiability” As we gain more wealth, as tech expands, we think of more things to do. Invention is the mother of necessity.
Issue will be how wealth is used. Think Saudi Arabia vs Norway. With respects to the use
Are we making progress in areas that are not making significant productivity gains.
“We all know more than we can tell.” we have gotten past this now.. The degree of uncertainty is truly unknown.
Humans are good on small data, based on our models of the world. We can make inferences based on disparate data.
The key challenges for executives, will be:
(1) shifting the training of employees from a focus on prediction-related skills to judgment-related ones;
(2) assessing the rate and direction of the adoption of AI technologies in order to properly time the shifting of workforce training (not too early, yet not too late); and
(3) developing management processes that build the most effective teams of judgment-focused humans and prediction-focused AI agents.
Professor Daniela Rus discusses the impact that robots will have on the workforce. She uses an example of applying an any-time optimal algorithm to match the taxi supply and demands in New York City, which reduces the number of taxis needed. She examines whether this could result in taxi drivers losing jobs. Professor Rus also explores the potential benefits of autonomous vehicles on mobility and quality of life. She discusses AI’s impact on fields such as healthcare, law, and education before talking about the current limitations of putting AI to work.
- NYC 14,000 taxis, MIT algorithm says only 3,000 taxis are required with a capacity of 4 passengers, satisfying 98% of demand within a 2.8min waiting time and a trip delay of 3.5min.
- Level 4 autonomy is here. autonomy in some environments… Level 5 has a way to go…
- Think of the walking stick being replaced.
- Machines are better medical/legal/teachers predictors, still need human judgment and emotional connections
- Machine learning has potential applications in so many fields, and medicine is a great example. Machines today can read more radiology scans in one day than a radiologist will see in a lifetime. So, a new AI-based approach was tasked with classifying radiology scans of lymph nodes as cancer or not cancer. The machine had 7.5% error, as compared to the 3.5% error of the human. But working together with a human, the machine and the human together achieved 0.5% error, which is a significant improvement over the state of the art.
- Gaps/limitations in AI in breadth and depth perception, reasoning, creativity, thinking….
- no universal tools
- Crunching data does not translate in to knowledge.
- complex calculations doe not produce autonomy.
- 99.99% is exponentially harder than 90% correct.
- Perception and action
- tasks with physical contact….
Professor Rus talks about jobs in terms of the tasks they entail. She sees a future partnership between people and machines in which each performs the elements of the job to which they are best suited: machines doing what’s easiest for machines and people focusing on the strategic tasks. She discusses two points of concern: productivity and job quality or wages. Lastly, she emphasizes lifelong learning and reinforces the idea of collaboration between people and computers.
- There will be a focus on tasks, not jobs.
- must be a lifelong learner…
Professor Malone asks Professor Frank Levy about the implications of AI on employment and the future of work.
- Politics needs to be a part of the story
- More people will be knocked out of mid-skill jobs
- Physical issues at the bottom, non-repetitive work at the top.
- Writing for MIT Sloan Management Review, H. James Wilson, Paul R. Daugherty, and Nicola Morini-Bianzino describe the jobs that AI will create and divide them into three new categories: trainers, explainers, and sustainers. Humans in these roles will work alongside machines, ensuring that machines are working in an effective and ethical manner.
- Read about which aspects of various jobs could be automated and which are less able to be automated, and gain further insight into the graph that Professor Rus used in Video 2 to illustrate automation across different activities in different sectors.
- Will a robot take your job? In January 2017, McKinsey Global Institute published a report estimating that by 2055 (give or take 20 years), around 50% of today’s work tasks could be automated. In this interactive graphic, you can input a job title or industry to find its automation potential.
- Ravin Jesuthasan and John Boudreau, in the Harvard Business Review, provide a four-step approach for thinking about how automation will affect job design.
- Read about the productivity benefits of automation along with its impact on various industries and implications for policymakers. (Access the report by clicking on the download link.)
- Have a look at five management strategies for getting the most from AI.
Over the longer term, a task-based view of work will be needed to make the best use of AI, to understand which tasks can be automated and which ones are better suited for people to do. New jobs will be created that are still unimagined. Institutions and society – the education system in particular – have a role to play to unlock the full potential of both people and machines in the future.
General ethical concerns surrounding AI
MIT Professor Iyad Rahwan highlights general ethical concerns about AI and explores why organizations should care about AI ethics. He describes how to balance the benefits and risks of AI, and he explains how people have been addressing these problems by using “a human in the loop.” He describes the regulation of human behavior and then examines the challenges involved with regulating the behavior of machines. He concludes by discussing preconditions to promote public trust in machines.
- Have discussed… technical aspects of AI, business value, strategic value, future of work
- Ethics: moral principles that govern behavior
- AI benefits:
- Better recommendations
- Safer Cars
- Better medical diagnosis
- AI Risk
- filter bubbles
- fake news
- unfair matching
- Need to put a “human in the loop”
- AI = prediction human = judgment?
- Society in the loop
- human in the loop, with a social contract.
- Regulatory forces, more than just the law…
- We have safety standards, liability laws, and consumer expectations
- Regulating AI is different
- not passive,
- have autonomy,
- have intentionality
- can adapt and learn
- cant certify at design time, as they will adapt and learn as they interact with the real world.
- will need to enforce the law, as they act
- Agency vs Experience
- machines don’t care about out norms
- How do you assign intentiality
- Need to understand emerging norms. What do we expect from AI, need to help adoption, while not over regulating.
Professor Rahwan provides a case study of the ethics of autonomous vehicles. He poses a scenario: what if an autonomous vehicle’s brakes become inoperable, and the car is heading toward a group of pedestrians who would be killed if the car hit them. The car has a choice to swerve and hit only one pedestrian rather than the group. Should the car swerve? Or, what if the car could swerve and avoid the pedestrians, but would thereby harm the occupant(s) in the car? At issue is that a machine would be making a moral decision. Professor Rahwan discusses approaches to this dilemma, including how different countries have started tackling such questions.
- Accidents by car: 1.2m -> 120k human to autonomous…. 90% accident are attributed to human error.
- what socially acceptable behavior
- social dilemma:
- I would not want to be sacrificed
- but everyone else should
- Our choices have externalities, we can’t be selfish
- People are less likely to purchase a car that will sacrifice them
- adaption and capacity
- theory of mind, mind perception
- bars on the front of the car. US ok, Europe no
- our new issue is that its a software decision
- open issue
- Germany created an autonomous car ethics commission
- Legal scholars
- Ethics experts
- Consumer protection groups
- Religious leaders
- avoidance of critical dilemma situation.
- should not be any discrimination…
- you can take total number of casualties in to account, not mandated
- person who does act on purpose, should not jeopardize people int he car.
- Check lout the Moral machine
- AI is a new kind of challenge, that we need to take seriously to address Law, Norms, Market, and Architectures.
2.3 Crowdsourced workers
Platforms such as Amazon’s Mechanical Turk (MTurk) and CrowdFlower, as well as vendor-managed systems like Clickworker, let companies hire contract workers to complete tasks, down to the level of microtasks that may only take a few seconds to complete. Read about the ethical issues that arise regarding the many crowdworkers whose low-paid, behind-the-scenes labor underlies many AI systems.
2.4 Biased algorithms
AI systems can excel at identifying patterns that let companies target specific customers more precisely. As you’ve seen throughout the program, this ability helps companies serve the unique needs of niche customers. But, sometimes this targeting can go awry. For example, Facebook’s algorithms enabled advertisers to reach self-described racists. Facebook’s COO, Sheryl Sandberg, publicly apologized for this “totally inappropriate” outcome and Facebook pledged to add 3,000 people to its 4,500-member team of employees to review and remove content that violates its community guidelines.
In another example, Microsoft set out to build a chatbot that could tweet like a teenager. Microsoft announced “Tay” on March 23, 2016, describing it as “an experiment in conversational understanding” and released it on Twitter. The idea was that the more Tay engaged in conversation with people on Twitter, the smarter it would become. Unfortunately, Tay learned all too well. As people sent racist, misogynist, anti-Semitic Tweets its way, Tay started responding in a similar tone, not simply repeating back statements, but creating new ones of its own in the same unfortunate vein. At first, Microsoft deleted the offensive statements, but, within 24 hours, shut down Tay to “make some adjustments.”
- Four researchers in the field of AI share their views, concerns, and possible solutions for reducing and avoiding societal risks associated with AI.
- Read about the differing views of Elon Musk and Mark Zuckerberg on the safety of AI.
- According to Sandra Wachter, Researcher in Data Ethics at the University of Oxford, although building systems that can detect bias is complicated, it is in principle possible and “is a responsibility that we as society should not shy away from.”
- In 1942, science fiction writer Isaac Asimov coined his Three Laws of Robotics in his short story, “Runaround.” The three laws are outlined in Figure 1.
Humans seamlessly integrate perception, cognition and action.
AI raises serious ethical concerns, as smart machines will make decisions that may have life-and-death implications. Many AI systems also rely on low-paid workers who labor behind the scenes. Finally, AI has the potential to exacerbate and amplify the negative qualities of humans. As a result, executives considering adoption of AI systems need to reflect thoughtfully on the ethical aspects of their choices.
- Professor Daniela Rus introduces the world of robots. She touches on computation in motion and different robot forms, and highlights some potential future applications of robots. She also explores advances in autonomous driving.
- most robots are built for specific tasks.
- Automated devices that perform physical tasks in the real world.
- Ways that humans and robots can be more effective together. trusted partners.
- Computation in motion
- Sensors to actuators
- Come in many different shapes/forms
- advances in HW and algorithms driving automated cars.
- Cheap Sensors and Actuators
- Body and brain
- Body: sensing
- Body: Computation
- Body: Actuation
- Body: communication
- Brain: Algorithms
- Professor Rus delves more deeply into the topic of autonomous vehicles. She describes the two different phases of autonomous systems and the architecture of the systems now being developed. This includes how autonomous vehicles rely on planning and control, localization, and feature vectors to get from a starting location to a destination. A feature vector is a list of numbers representing an object’s important characteristics, such as size, shape, color, and brightness. She also explains parallel autonomy: when a system corrects a driver’s errors by making small adjustments, such as turning the wheel to ensure safety.
- Two phases for autonomous vehicles
- Phase 1 – Maps the environment, map is a feature space.
- Phase 2 – Plans a path and executes from start to destination, perception
- detect obstacles
- perception is performed by looking for features, to identify uniquely most locations in the world
- curvatures of curbs
- textures of buildings surrounding roads
- From normals, you can extract out features to form curves, etc..
- have a virtual bumper, to give enough time to create an alternative movement plan
- import traffic rules, but know when to break the rules per Vienna convention
- IMU (Inertial measurement unit)
- Series Autonomy: either human or computer is in control. Think Google
- Parallel Autonomy: human in control, but “guardian agent” is there. A Shared Controller works between human and Series Autonomy system.
- Professor Rus examines the applications of autonomous vehicles by using two examples: an autonomous vehicle trial in Singapore and the potential impact of driverless taxis in New York City. She also discusses technological and policy challenges associated with the adoption of autonomous vehicles.
- Cars and public transportation system are connected.
- Algorithm Optimal path for single vehicle, pooling. system implemented using taxi data in NYC
- autonomy and ride sharing can transform in to a utility, reduce congestion, decrease pollution, improve quality of life.
- what’s hard
- human gestures
- bad weather, poor visibility
- high speeds
- Computer vision is getting better, but perception is still lacking
- Policy is lagging.
- dedicate lanes
- parallel autonomy
- automated logisitics/highways
- Professor Malone interviews Professor Rus to discuss four topics:
- Why the robots in use today don’t walk around like robots in sci-fi movies?
- Important to think about capabilities of body, relative to task.
- legged robots are hard in the real world. Where to place the next leg. Boston Scientific.
- Robot would need extraordinary perceptions and agility of motion.
- What is easy and what is hard in robotics?
- mobility is easier than manipulation.
- will we see a flying car first, or Level 5 vehicle?
- manipulation, but soft robotics are advancing… will have greater compliance like human manipulation.
- Must be able top figure things out on their own.
- better at interacting with people
- making new robot bodies faster.
- How people and robots are likely to interact?
- Working side by side.
- Shared control of certain tasks. they should be able to override each other.
- leveraging each others best assets, being better together.
- The kinds of robots, besides self-driving vehicles, that are mostly likely to be used in business in coming years?
- that a machine and human working together will be better than individually.
Business applications of robotics
- Professor Malone discusses the three other main ways robotics are being used today:
- in factories…
- to perform miscellaneous physical services.
- deliver, manipulate, locate physical objects
- Sensing, Computation, and Actuation
- Factory robotics:
- Acorn Sales Company, a small manufacturing company, talk about how they have deployed a robot called Sawyer, built by Rethink Robotics, to improve productivity.
- Gained acceptance from employees to do the dull and dirty tasks of lifting and positioning material onto conveyer belts and sawing and drilling wood.
- Acorn uses Sawyer as part of its cost leadership strategy to be able to keep prices in line with those of competitors. Goals:
- Less reliant on other suppliers
- increase quality
- keep prices in line with competitors
- business continuity
- Baxter can learn from EEG sensor connected humans who provide reinforcement through thought.
- Warehouse robotics:
- Kiva’s robots have revolutionized operations at Amazon’s warehouses. To achieve efficiency gains, Amazon re-engineered its warehouse processes based on Kiva’s capabilities, rather than trying to fit the machines into existing processes.
- The use of robots in warehouses enables productivity improvements, thus supporting a company’s cost leadership strategy, particularly in times of labor shortages. Robots can lift and move heavier loads, bringing items to easy reach of human workers who then select the products to fulfill each customer order.
- From a competitive advantage standpoint, it’s interesting to note that Amazon acquired Kiva Systems, renamed it Amazon Robotics, and then chose not to renew the contracts Kiva had with other companies, thereby confining the technology to Amazon and keeping it out of the hands of competitors.
- Performing miscellaneous physical services:
- Savioke’s robots, Relay and Botlr, are armless robots used in hotels to transport amenities autonomously to guests’ rooms. In the next two videos, Professor Malone interviews Savioke’s CTO and Chief Robot Whisperer, Dr. Tessa Lau, to find out more about Savioke’s robots, their appeal to customers, and their implications for business strategy.
- Dr. Lau provides an overview of the role of Savioke’s Relay robot in hotels. She talks about how Relay moves around hotels to deliver items to hotel guests, and she discusses the benefits of using a robot like Relay in the hospitality industry. She also mentions how Savioke’s robots are used in other industries. Professor Malone and Dr. Lau discuss how Savioke has created robots that navigate through human spaces and the company’s focus on human-robot interaction design.
- deliver robots for indoor space. Relay is for hospitality. controls elevator through wifi…
- Highest turnover in the US… Robots provide consistency, at a lower price point deliver less than 5 min predictably
- FedEx uses 7 robots, repair facility, relay brings parts to them as they need them. (like Kiva)
- focused on human spaces. Robots that are safe for people. Makes Relay very aware of surroundings.
- Human-Robot interaction design.
- Asks for help,
- Human has empathy for robot.
- Anthropomorphizing robots
- By making him cute and approachable increase adoption, actually protects him in the real world…
- Professor Malone asks Dr. Lau to expand on the AI technology that drives the Relay robot, and they discuss the qualities that Relay has in contrast to a human employee. Dr. Lau highlights some of the limitations and challenges that Relay still faces, and she concludes by providing advice to people who are interested in using robotics.
- primary technology is robot navigation/path finding. How do I get from one point to another safely, within constraints and obstacle….to construct a plan….use AI to actively replan.
- biggest competitor is human labor.. but brings reliability, consistency, automation, and predictably.
- How to make relay safe in some environments, (e.g. escalators, stairs,
- limited to wheelchairs can go. ADA creates standards that robots can use.
- Still needs to be programmed to do the task. The more you can cast your problem in the eyes of repetition. Maybe processes need to change to become structed for repetition, you don’t have to think.
- Key points and strategy
- Robots that operate in the same environments as humans must have enough understanding of their surroundings and of the humans in them, so they can navigate those environments safely. This requires machine vision (like autonomous vehicles have), as well as “path finding” (the ability to get from Point A to Point B even though there may be many ways to get there). The robot must also be able to reprogram if the environment changes, such as if a person steps into its path or the robot ends up on the wrong floor.
- One reason to give a robot a human “face” and “personality” is to encourage empathy for the robot. Just as babies have big eyes to trigger an emotional connection, a robot that can create an emotional connection will be more accepted and be perceived as trustworthy by people. Savioke designs its Relay robots with attention to making them “polite” in interactions with humans.
- A delivery robot can perform certain tasks less expensively than human labor can, supporting a cost-leadership strategy in industries such as hospitality that have high turnover requiring training of new workers. However, there are many tasks robots cannot do, and thus they cannot replace human workers entirely. A delivery robot can help to ensure consistency of service, thereby also supporting a differentiation strategy. For example, a hotel brand could advertise that a guest, staying in any of its properties, could expect room service delivery within 5 minutes, because that is the typical time it takes a delivery robot like Relay to fulfill a housekeeping request.
- Robots are currently deployed in factories and warehouses, and sometimes behind fences or in restricted areas to keep them separate from humans, for safety reasons. But as robots become increasingly able to distinguish humans from other objects, they are becoming safe enough to operate in the same physical spaces as humans, such as in hotels and restaurants. Organizations that automate manual tasks so that robots can perform them can potentially reduce errors and lower costs while improving efficiency and quality, often all at once. In many settings, the most successful outcomes have been achieved through robots and people working together, each doing the tasks that are easiest for them to do.
Module 3 of 6 of MIT CSAIL AI Implications & Strategy: Natural language processing in business – Notes
Professor Regina Barzilay introduces the field of natural language processing (NLP). Professor Barzilay discusses what it means for machines to understand something, and she delves into the history of conversational devices. She explains natural language processing tasks that are considered solved (such as spam detection), tasks where progress is being made (such as sentiment analysis), and tasks that are still difficult for machines (such as question-answering systems).
- what they can do and what they cant
- intuition of what’s possible
- What is NLP?
- Natural language (e.g. English, Spanish, Russian..)
- Natural language processing (NLP) is a field of computer science, artificial intelligence concerned with the interactions between computers and human (natural) languages, and, in particular, concerned with programming computers to fruitfully process large natural language data.
- Language technology
- SPAM DETECTION
- PARTS OF SPEECH TAGGING
- NAMED ENTITY
- Making Progress
- SENTIMENT ANALYSIS
- CO-REFERENCE RESOLUTION
- Word sense disambiguation
- Machine translation
- Still Really Hard
- Question Answering
- Syntactic parsing
- Close to solved problem
- Machine translation (MT)
- Think “Google Translate”
- You need to train the system to get the results you are looking for….
- Question answering (Q&A)
- Think Watson
- machines have a better ability to search enourmouse amounts of inforkmation, quickly..
- However, doing QA based on reading a paragraph still has challenges
- Text summarization
- Again, machines trained on creating summaries can do a good job of creating them.
- What NLP cant do automatically
- Being able to take domain specific content and pulling out what’s important for summarization is still outside the realm of possibility for NLP.
Professor Barzilay describes how NLP is generally applied. She illustrates how machines can perform well on some tasks that are considered difficult for humans and perform poorly on some tasks that are considered easy for humans. She then explores the challenges associated with NLP by describing different levels of language ambiguity.
- NLP in industry
- Information extraction
- Machine translation
- text generation
- Sentiment analysis
- Challenges for NLP
- Anne Hathaway vs Berkshire Hathaway….
- Difficulties with ambiguities, examples:
- Ambiguity for NLP
- Need to dis-disambiguate to get understanding. this is difficult.
- Anaphora, where say a pronouns can co-refers to some other discourse entity.
- What is a word? is it just a space, this changes across languages.
Professor Barzilay explains how NLP has evolved by looking at the different approaches to making machines understand natural language. She describes a statistical approach versus a symbolic approach, as well as recent breakthroughs in the field. Professor Barzilay then talks about an approach that uses supervised classifiers and discusses modern techniques related to deep learning that are currently dominating the NLP field.
- Knowledge bottleneck in NLP
- Needs to know grammar and world around it…
- symbolic approach: all coded in…
- Statistical approach: give language samples…
- Noam Chomsky – MIT prof, pioneer
- “Colorless green ideas sleep furiously.” vs.
- “Furiously sleep ideas green colorless”.
- he felt…statistical approach would not find the difference…
- “Whenever I fire a linguist, our system performance improves.” Jelenick 1998
- Symbolic Era
- SHRDLU, tried to encode everything in to a machine.
- however, even the smallest of domains have too much information to capture every circumstance.
- Statistical Era
- DARPA developed a tree bank… took sentence and developed syntactic mapping..
- This developed robust parsings
- Determiner placement
- Where to put here and there… hard for non English speakers,
- Supervised learning in NLP
- What if you model with supervised classifications?
- Need to represent data as a feature vectors with + or – decisions… Supervised
- Deep learning
- You can use to generate the feature vectors.
- Deep learning has enabled a revolution in NLP, and that we will be able to achieve what Turing believed in a short period of time.
Professor Barzilay tells Professor Malone that having training data of a good quality is a key enabler that allows NLP systems to work effectively. They discuss NLP as a branch of machine learning and explore the use of human-powered annotation services, staffed by workers from Amazon’s Mechanical Turk, to augment the machine learning engine.
- Important NLP considerations
- What kind of tolerance does customers have to system mistakes? (filtering links on Search, Google translation being good enough), not meant for legal docs, medical on its own..
- What kind of training data do you have? The training data needs to align on the problem you are solving for. Creating training examples can be very expensive, but is very important.
- NLP as a branch of machine learning
- takes in unique properties of sentence structure, etc.. unique models like sequence to sequence methods.
- Understanding features
- It’s important that humans flag what’s important, where there may be correlations…
- need to give knowledge as inputs to the models.
- Annotation services
- You can use services like mechanical turk for annotation, but probably not for domain expertise. You use them to label the data.
- ML expert will determine how to code the data to maximize learning…
Professor Barzilay mentions two ways that natural language can be generated and discusses with Professor Malone the challenges involved in trying to get machines to use deeper, more generalized kinds of reasoning.
- Natural Language Generation, two methods
- Completely from scratch, training with a large amount of data.
- Template generation, many responses, with blanks, grab the right template, and plug in the data.
- Current limits in reasoning
- Systems are trying to learn from observations (from watching & listening) to draw out inferences.
- The future is learning through human observation… more abstract learning.
- You want to be able to just tell the machine what you want…
- Today you need a lot of examples….
Casebook: Business applications of NLP
An enormous amount of data in the form of unstructured human language is currently being generated every minute through a variety of channels, such as news articles and blog posts, tweets, and posts on platforms like WhatsApp and Facebook. Business communications include those between people working together, and online interactions between companies and their customers. This module focuses on how companies can deploy natural language processing (NLP) to derive value both from understanding some of this vast amount of unstructured language and from generating natural language responses to it.
Professor Thomas Malone talks about three things that artificial intelligence programs can do with natural languages. The three functions of natural language processing are then explored through the examples that follow the video.
- Understand Text/Speech
- Generate Text/Speech
- Converse in Text/Speech
- Natural language understanding
- Call centers
- balancing the trade off between customser services with lower costs
- A.I. can triage callers questions, and who may be best to handle the call..
- Three components
- Automatic speech recognition
- natural language processing, meaning
- information retrieval
- Is this a factoid or more involved response?
- The things it can do well, are those things it has seen before.
- System must be trained for the specific task/problem.
- der questions… Those things it hasn’t seen before. Multiple responses are generate, however it’s the one with the highest confidence, above a threshold that is typically used; otherwise it goes to a human.
- Key points and strategy
- NLP is being used in call centers in service of two potential strategies: cost or differentiation. Text- or voice-based chatbots can do some of the work previously performed by human customer service representatives, making it possible to reduce the number of people needed, thereby lowering costs. Most companies installing customer service chatbots, however, are doing so in service of a differentiation strategy. In this case, the chatbots take care of routine matters while more complex requests are transferred to human representatives, leading to a higher quality of service.
- Understanding documents
- Need the representation of the language first.
- Repetitive processes are easy. Document classification in discovery before a trial.
- Keywords not enough
- Machine learning let lawyers classify sample, then use that to classify documents. confidence .9 take, .4 or less reject, between human review.
- Contract management
- in due diligence… Look for clauses, written different ways (cancel contract if firm is acquired)
- Key points and strategy
- As Professor Levy explained, NLP can be used in the legal discovery process to identify responsive documents that must be turned over to the opposing party (or that clearly do not need to be turned over). The NLP system gives its confidence rating on each document, and the documents that have a high confidence rating can be automatically turned over. If the NLP system is uncertain about a document, a human can step in to make the final decision. By automating routine tasks, NLP can support a lower-cost strategy. But more often, time is freed up for higher-value tasks, supporting a law firm’s differentiation strategy to provide a higher level of service to its clients.
- IBM Watson
- PDF text vs image and being able to extract out letters and words to meaning (concepts…) from either.
- Tool for compare and comply by extract meaning between two documents, then comparing then.
- Associate words to higher categories or concepts, can link with Wikipedia entry to get more evidence, this starts top create layers.
- knowledge graph starts to build out insights and understanding, you don’t have to learn from scratch.
- Can takes different modules (skills) access via APIs to build solutions.
- More work needs to be done in learning and reasoning.
- Reasoning is an important issue in AI… Lots of academic and industry work is happening here.
- Natural language generation (NLG)
- Narrative Science
- Take a look at data, and humanize it. From data to a narrative (description), and a piece of advise (prescription)
- Quill is the technology
- This is a story of focus.
- Key points and strategy
- Companies can use NLG to support several of Porter’s generic strategies. NLG can support a cost-leadership strategy by producing text much more inexpensively and quickly than human writers. On the other hand, NLG can also support a differentiation strategy by freeing up writers to focus on higher-value tasks.
- Interacting in natural language
- intelligent agent, that can schedule agents
- Key points and strategy
- Virtual assistants like x.ai can be used in a low-cost strategy, to reduce costs of customer interaction, or in a differentiation strategy, by freeing up staff to do higher-value work, thereby improving quality. However, some of what x.ai is doing today is still done by people.
- Amazon’s Alexa is a voice-based assistant that shows what is possible with NLP. Users can speak out loud into Amazon’s Echo speaker to ask Alexa to perform any number of a growing number of tasks, such as to tell a joke, play a song, give the weather forecast, lead an exercise routine, or order products on Amazon. Google, Microsoft, Apple, and China’s Baidu also have voice-based assistants of this kind. Amazon can also mine the queries that Alexa cannot yet understand or satisfy. If enough people ask for sports scores, for example, it is likely that this kind of functionality would be popular, and Amazon may decide to build it. Amazon also lets other companies such as Uber, Fitbit, 1-800-Flowers, and Campbell Soup build apps on the Alexa platform. As of June 2017, Alexa had 15,000 such apps, although many of them remain quite simple.
- Interacting in natural language (continued)
- Key points and strategy
- As this example shows, NLP can be used to support a strategy of differentiation, whereby companies that build apps on Alexa make it easier and more convenient for their customers, thus increasing the quality of the goods or services they provide.
- It’s a method called wavenet, discovered by researchers at Google’s DeepMind and published as a research paper in September 2016. The method uses a particular type of neural-network architecture to create sound, and is said to represent a significant leap forward in artificial-voice technology. Voysis
- As a technology, natural language processing is making daily life easier by automating tasks, sorting through data with a level of speed and accuracy that humans are not capable of, and making connections between data to enhance and personalize the online experience. It is being applied to a wide range of business problems to deliver tangible business value and will continue to transform how people live and work.
MIT AI M3U1 Video 1 Transcript.pdfMIT AI M3U1 Video 2 Transcript.pdfMIT AI M3U1 Video 3 Transcript.pdfMIT AI M3U1 Video 4 Transcript.pdfMIT AI M3U1 Video 5 Transcript.pdfMIT AI M3U2 Casebook Video 1 Transcript.pdfMIT AI M3U2 Casebook Video 2 Transcript.pdfMIT AI M3U2 Casebook Video 3 Transcript.pdfMIT AI_M3 U2 Casebook.pdfMIT AI_M3 U3 Assignment.docx
Module 2: Machine learning in business
Unit 1: Key features of machine learning
Machine learning as a discipline seeks to design, understand, and use computer programs that learn from experience (i.e., data), without being explicitly programmed for specific modeling, prediction, or control tasks. In his interview with Professor Malone, he suggested three requirements for using machine learning:
- The problem can be formulated as a machine learning problem.
- There is much relevant data available that could be used by machine learning algorithms.
- The system has enough regularity in it that there are patterns to be learned.
Understanding machine learning
Professor Jaakkola provides a brief definition of machine learning, and discusses how it can be used to solve prediction problems. He also gives examples of the types of prediction problems that can be addressed, and touches on supervised learning and convolutional neural networks. He then highlights the recent progress made in the machine learning field, discussing deep learning architectures, with DeepMind AlphaGo used as an example.
- programmers use tools to abstract out the complexity of writing machine code
- what about creating general learning program that learn from experience, data
- Tyopes of Prediction problems
- future events
- predict properties not know yet.
- formulate ml problem, by learning from examples
- Labeled data is a group of samples that have been tagged with one or more labels. Labeling typically takes a set of unlabeled data and augments each piece of that unlabeled data with meaningful tags that are informative. For example, labels might be indicate whether a photo contains a horse or a cow, which words were uttered in an audio recording, what type of action is being performed in a video, what the topic of a news article is, what the overall sentiment of a tweet is, whether the dot in an x-ray is a tumor, etc. Labels can be obtained by asking humans to make judgments about a given piece of unlabeled data (e.g., “Does this photo contain a horse or a cow?”), and are significantly more expensive to obtain than the raw unlabeled data. After obtaining a labeled dataset, machine learning models can be applied to the data so that new unlabeled data can be presented to the model and a likely label can be guessed or predicted for that piece of unlabeled data.
Supervised learning – Supervised: given explicitly input examples and target label, that I wish to predict: … is the machine learning task of inferring a function from labeled training data. The training data consist of a set of training examples. In supervised learning, each example is a pair consisting of an input object (typically a vector) and a desired output value (also called the supervisory signal). A supervised learning algorithm analyzes the training data and produces an inferred function, which can be used for mapping new examples. An optimal scenario will allow for the algorithm to correctly determine the class labels for unseen instances. This requires the learning algorithm to generalize from the training data to unseen situations in a “reasonable” way (see inductive bias).
- Typically had to program in detail what you wanted a machine to do, that is then translated in to lower level code. But now instead, you can write a general program that included instruction to learn from experience (i.e. data).. then teach them what to do by giving them more data and direction….
Formulate by learning from examples.
- Training set: Label examples of movies I do and don’t like +1 / -1
- Translate in to feature vector, based on questions we would ask…
- Genre, famous lead, etc? Then calculate in to a binary feature vector
- Test set: Then I go through the same process for movies that I have not seen.
- Training set give us clues, to predict the label in the test set. Then the algorithm classifies new movies presented in to like and dislike.
In computer vision
- Learn from idneitifyed featured in multiple passes.
- Layeers of transfomariont, then you get a deep model … Convulution Neural networks
- Movie recommendation
- Take descriptions and translate in to a feature vector
- Ask questions about the movies, then calculate those answers in to a vector… What is the genre, famous lead, etc… then calculate in to a binary vector….
- Then calculate the same vector for movies you have note seen…
- In the training set you have labeled vectors; in the teat set you just have the vectors that you need to provide the label for…
- Move into a geometric form and put the vectors as a point in space, based on labels.
- problem, how to classify test set, training set gives me clues to classification of data generally.
- You can divide data in to halves of -/+ then bring back test set and classify them in to the two halves, based on the learning form the training set.
- Convolution neural network,
- Classify images in to content categories
- progressive learning of what an Item is by looking at a fraction (small features) of the pixels then gradually taking in more and more of the image.
- Recent progress
- 1- Error has gone down by 50% to 100% per year… 2012 -> 2014…
- 2 – The red represents deep learning approaches, and you can see how they have taken over computer vision
- We also see advancements in machine translations, captures semantics, etc… You are giving an example of the correct behavior, then you are trying to automate the process of finding the solutions
- Learning to act: Playing Games
- GO – Looked at game board and learned to match it to what actions you could take, instead of thinking about all of the possible actions upfront. it did this by watching human experts play. However, you could probably do better by watching a computer play. Deepmind alpha go
Professor Jaakkola provides four main reasons for the recent advances made in machine learning, as follows:
- The accumulation of huge amounts of data
- Advances in computational power
- The growing complexity of models
- Large models are easier to train
- The new possibilities created by deep learning architectures
- Flexible neural “lego pieces”
- Common Vector representaiton
- Allows you to take an image and generate a word, or sentence, or take a word or sentence and generate an image… Because this model has been put in to vectors that are cross referenced. Allows the transfer from one domain to another in a very simple way…
- Recurrent neural networks, takes known information and new information, to derive a new vector representation. and see how it functions for the classification task you will want to use it for.
- Google translate works in this way…
- Takes a lot of data to get examples of behavior.
- Amount of data: What data i have prior to trying to solve a problem… What additional data do I need.
- Types of machine learning:
- Unsupervised learning is a type of machine learning algorithm used to draw inferences from datasets consisting of input data without labeled responses. The most common unsupervised learning method is cluster analysis, which is used for exploratory data analysis to find hidden patterns or grouping in data.
- Semi-supervised learning is a class of supervised learning tasks and techniques that also make use of unlabeled data for training – typically a small amount of labeled data with a large amount of unlabeled data. Semi-supervised learning falls between unsupervised learning (without any labeled training data) and supervised learning (with completely labeled training data). Many machine-learning researchers have found that unlabeled data, when used in conjunction with a small amount of labeled data, can produce considerable improvement in learning accuracy.
- Active learning is a special case of semi-supervised machine learning in which a learning algorithm is able to interactively query the user (or some other information source) to obtain the desired outputs at new data points.  In statistics literature it is sometimes also called optimal experimental design.  There are situations in which unlabeled data is abundant but manually labeling is expensive. In such a scenario, learning algorithms can actively query the user/teacher for labels. This type of iterative supervised learning is called active learning.
- Transfer learning or inductive transfer is a research problem in machine learning that focuses on storing knowledge gained while solving one problem and applying it to a different but related problem. For example, knowledge gained while learning to recognize cars could apply when trying to recognize trucks. This area of research bears some relation to the long history of psychological literature on transfer of learning, although formal ties between the two fields are limited.
- Interpretable modeling
- deep learning is pretty opaque, so we need to make them interpretable… highlight the rationale to the solution/predictions. This will help us guide them.
Professor Jaakkola explains, at a high level, the difference between
Shallow learning and
- say in image prediction (recognition) it would be taking all pixels and bouncing off of a simple classification model to see if you can determine what it is… based on all available pixels. Pixels -> Cat or human
- “C” or “D”
Deep learning architecture
- layers of processing, features, combination of features breakdown of data in to discrete parts to understand them, then bringing together to provide the prediction. Pixels -> arm -> hands -> fingers -> eyes -> edges -> wiskers -> etc etc etc etc ->humam or cat
- “Dog” or “Car”
and talks about
neural networks and
convolutional neural networks.
- detectors for each area of an image… Spatial analysis
The two professors discuss what is currently both easy and difficult to do with machine learning algorithms. Professor Jaakkola poses three questions to help you decide whether machine learning is possible as an application to solve a problem.
- Can you formulate in to a machine learning problem?
- You can use supervised learning to predict many things. Provide training data and expected results, and this can be used to find answers in target data sets.
- Do I already have tools available to address that formulation?
- Classifying images
- Interpretating natural language
- Business decision (stock picking)
- Is it likely to work well. as a solution to the problem?
- With formulation and tools, still unable to develop high confidence results…
- Example: Planetary motion can be chaotic…
They talk about the differences between
- labeled data with target outcomes.
- Trying to understand based on observation/how things work.. what are the regular structures… Then when you get feedback, it increasing the learning.
- used to learn to action, based on changes to the sate of the world, you need to act differently…
as well as the amount of training data or examples that machines need to make predictions.
- Supervised: Do you have historical data, actions and outcomes? featurize data.. attributes being tracked.
- You can use unsupervised if you dont have a lot of data, just to start.
- If you have feature representation exists, then very little data one example will allow fpor a prediction, but that is based on a lot of data to train the system
- Starting from scratch – Tabula rasa refers to the epistemological idea that individuals are born without built-in mental content and that therefore all knowledge comes from experience or perception.
- if it’s easy (high correlation) could be few 100 examples…
- if it’s hard (low correlation) could be much more…
- exactly how much data is based on the complexity of the formulation problem, and can be assessed by a data scientist.E
machine learning systems currently can’t properly explain how they came up with an answer and whether this will change in the future.
- Active area of research. Interpretability. How and why a decision was made.
- No commercially available solutions at this time. Possibly next year or two. Need will drive this… Think medical fields.
Lastly, he explains how formulation is key to understanding machine learning.
- Formulation is the key. That is… In this case, this is what I want…and here are illustration of what I want. to then run through ML.
- What are the inputs, and what are the outputs you are looking for…
- Recognize machine learning problems all over the place…
Three requirements for using machine learning:
- The problem can be formulated as a machine learning problem. Formulation is the key. That is… In this case, this is what I want…and here are illustration of what I want. to then run through ML.
- There is much relevant data available that could be used by machine learning algorithms.
- The system has enough regularity in it that there are patterns to be learned.
Unit 2: Business applications of machine learning [± 2 hour 30 minutes]
An executive’s guide to machine learning:\
In an article from McKinsey & Company, machine learning is explored from an executive’s perspective. The article discusses how organizations are using machine learning for insights, the importance of strategy in getting started with machine learning, and the role of senior executives in leading such initiatives.
- Perceiving large amounts of data from sensors in the world, and learning what to recognize what there.
- Recognizing movement, sounds, temps, light, vibrations, faces, retina, finger-prints, scenes for autonomous cars…
- Key points and strategy
- Applications like Shoegazer could be used to reduce marketing costs (customers point the app to a pair of shoes and click to buy a pair for themselves) or to support the development of new features for shoes (customers use the app to show features they like). This kind of machine learning can be used not only for images, but also for sounds (for example, Shazam is a mobile phone app that can identify songs)
- Key points and strategy
- People play a role in improving the accuracy of the scene analysis by identifying the objects that the machines have flagged as unknown, and then feed those images into the training set for the machine, so that future versions become smarter. Mobileye’s customers use the company’s AI technology to differentiate their products.
- What will happen in the future.financial fraud, disease, mechanical failure, crop yields.
- can customize these models by firm by customer base to get very customized based on the situation.
- Key points and strategy
- As Professor Lo explained, machine learning algorithms examine different parameters from those which traditional credit-scoring models examine. The insights gained from the machine learning techniques provide banks with a new lens that enables banks to predict consumer delinquency with a higher, finer-grained accuracy. Using this type of machine learning application, banks can pursue a cost reduction strategy that reduces their losses from non-payment.
- As this example shows, AXA uses machine learning to predict which customers are most likely to cause accidents that would cost AXA US$10,000 or more. Such predictions support a focus-based strategy, because they enable AXA to write policies only for lower-risk customers.
- PayPal uses machine learning to support a low-cost strategy of being efficient in customer service while avoiding fraud and customer inconvenience. Specifically, PayPal uses machine learning to determine whether a visitor to the site is a trustworthy customer. If the user seems suspicious, the system will ask for additional verification. The combination of neural networks, vast quantities of data, and deep learning have greatly improved fraud detection, but the machines cannot do it alone; people need to decide which data is relevant for the machine to use.
Personalizing product offerings
- Key points and strategy
- Pursuing a strategy of differentiation, Netflix combines machine learning with human curation to tailor its offerings precisely to the tastes of each individual customer.
- Like Netflix, Stitch Fix uses machine learning to pursue its strategy of differentiation, by personalizing its product offerings for each customer. Furthermore, human stylists and the company’s algorithms work hand in hand. The algorithms augment the human stylists’ productivity by doing tedious tasks such as matching client measurements to different brands and products. The stylists meanwhile read the personal notes that customers have sent and analyze their Pinterest boards to determine each customer’s nuances of style and taste. Stitch Fix’s approach illustrates three lessons about how to combine human expertise with AI systems.
- It’s important to keep humans in the business-process loop; machines can’t do it alone.
- Companies can use machines to supercharge the productivity and effectiveness of workers in unprecedented ways. And
- Various machine-learning techniques should be combined to effectively identify insights and foster innovation.
Improving product performance using better predictions
- Prof. Randall Davis… Digital Cognition Technologies (DTCclock)
- Analyze the product and the process, to screen people for cognitive problems.
- Reliability – test always done the same way from person to person.
- Detailed and informative measure – accuracy of screener
- Early indications of cognitive problems
- You can do the same in other aspect of business. Know it before it breaks
- Good sensors,
- good human interpreters of sensors, and
- good machine learning tools
- Not just about the technology, it’s an interplay between human and computers
- Humans: Development of initial system, and then dealing with exceptiopns.
- Analyze pen-strokes for classification, then take in geometric and temporal data for assessment, looking for infirm and normal.
- sketch understanding
- recognize most indicative features of data gathered.
- Buyer considerations for AI applications
- Does it work, show me tests, and performance, show me failures.
- Do you have the right kind of expertise in-house to develop, improve, and maintain?
- Keep the no magic principal…
- Key points and strategy
- As Professors Malone and Davis discussed, the machine learning aspect of the DCTclock can be generalized to many different applications of a screening test to provide early indication of a potential problem. In the case of the DCTclock, the machine looks for patterns that indicate cognitive impairment. The machine was trained using an expert’s knowledge (a neuroscientist) about the features that are early indicators of potential problems. The machine learned to find the patterns that indicate the early onset of impairment. Human and computer work together in that the machine can suggest which patients a clinician may want to watch or follow up with, and the clinician then provides the diagnosis and treatment therapy. The same process applies to the early screening of any kind of problem, such as a factory machine or a jet engine. Computers can alert people to an engine that may need repair, for example. The people then use their expertise to pinpoint the cause of the upcoming problem and take preventative action. Systems like these can support a differentiation strategy by allowing for more accurate predictions about future problems.
- Professor Alex Pentland from MIT talks about honest signals and how they can be used to interpret human intention and emotion. He talks about the “second language” that people use, consisting of body language and signaling behavior, as a result of some of our basic neural processes. Professor Pentland uses the example of Cogito’s software, which has been used to analyze conversations between call center agents and clients in real time. Computers and people work together, as the machine learning system interprets whether the conversation is going well or not, and then provides feedback to the agent to adjust their behavior for a more engaging and effective conversation.
- Listening to the non-linguistic aspect of a conversation to make all center reps more effective.
- Think about the signals we ca not see/hear, as we focus just on words.
- Cogito…slow down, stop talking, redirect…
- increase customer engagement
- more effective communication
- less conflict
- lower call center turnover
- Key points and strategy
- Professor Pentland’s research has found that “unspoken” language – the nonlinguistic signaling of interest, attention, dominance, and so on – accounts for 40–50% of the outcome of a conversation. People tend to focus consciously on words, so it is more difficult for them to tap into these signals, but computers can help them out. Professor Pentland used supervised machine learning in creating Cogito to “listen” to these unspoken features and predict how the conversation between a customer service representative and a customer is going. Cogito’s software supports a differentiation strategy by facilitating higher-quality customer service interactions by letting representatives know if the customer is paying attention or is getting angry, for example.
MIT AI M2U1 Video 1 Transcript.pdfMIT AI M2U1 Video 2 Transcript.pdfMIT AI M2U1 Video 3 Transcript.pdfMIT AI M2U1 Video 4 Transcript.pdfMIT AI M2U2 Casebook Video 1 Transcript.pdfMIT AI M2U2 Casebook Video 2 Transcript.pdfMIT AI M2U2 Casebook Video 3 Transcript.pdfMIT AI M2U2 Casebook Video 4 Transcript.pdfMIT AI_M2 U2 Casebook.pdf
Module 1 of 6 of MIT CSAIL AI Implications & Strategy: An introduction to artificial intelligence – Notes
Unit 1: Outcomes of this program
- It not about smarter machines, it’s about smarter organizations…
- Collective Intelligence: Groups of individuals working together, in ways that seem intelligent…
- In tech think of Google, or Wikipedia, Technology enabled collective intelligence
- Question: How can people and computers be connected, so that collectively they act more intelligently than any person or group has done before.
- Connecting people together in new ways so they can act more intelligent as a group
- Connecting people to computers that have more Artificial intelligence
- After the course, I will know…
- more about AI in business
- opportunities for AI in busienss
- ideas and concrete plan as to how MY org can use AI
- Will understand generally about AI in the world.
- If you are afraid of AI, the hope is that you will not be at the end.
Unit 2: An overview of AI: definition, history, and current state
- Defining AI
- The term “artificial intelligence” is not easy to define. The word “artificial” is more straightforward, meaning something that doesn’t occur naturally. By contrast, “intelligence” has been defined in many ways. One good definition, by the psychologist Howard Gardner, focuses on problem-solving: “Intelligence is the ability to solve problems, or to create products, that are valued within one or more cultural settings” (Gardner 1983). Informally, people sometimes use the term “artificial intelligence” to mean only those activities that are hard for computers to do (like understanding English) as opposed to simpler activities computers routinely do today (like accounting).
- An important distinction in the field of AI is between “narrow AI” and “general AI”. Narrow AI is defined as “a machine-based system designed to address a specific problem (such as playing Go or chess)” (Kiron 2017). By contrast, general AI refers to machines with the ability to solve many different types of problems on their own, like humans can. To date, all applications of AI are examples of narrow AI. Although general AI is currently a hot research topic, it is still likely decades away from true realization.
- For the purposes of this program,
- Professor Malone’s intuitive definition of AI is that it is: “AI is…machines acting in ways that seem intelligent”.
- Professor Patrick Winston’s more formal definition is: AI is about the architectures that deploy methods enabled by constraints exposed by representations that support models of thinking, perception, and action. And of course, it’s not just about doing, it’s also about learning to do.
- Professor Patrick Winston on the history of AI
- Professor Malone’s intuitive definition of AI is that it is “machines acting in ways that seem intelligent”.
- Computation isn’t just changing somethings, it’s changing everything
- “Founders” in the field of artificial intelligence
- Alan Turing, 1950 paper.. “Turing test”.. 5 min computer of person… trying to deal with objections of humans, vs just a test of intelligence. – We can do this…
- Marvin Minsky and the suitcase of words for AI… “It’s a suitcase term.” Minsky’s paper “Steps towards artificial intelligence 1961” – What to do to get there…
- AI’s 1st Wave… 1960’s
- James Slaegal, MIT work in symbolic expressions. All about Problem Reduction, or breaking problems in to simpler, and simpler problems…
- If you get the representation right, you’re also most done.
- Professor Patrick Winston “AI is about models of thiking, preception, and action…”
- model … behaves like the real things…
- representations .. support models… set of conventions for describing situations.
- representations expose constraints
- methods to deals with constraints
- Architectures for the methods.
- AI’s 2nd wave – mid 1970
- Ed Shirtliffe, Stanford , MYCIN – Diagnose infectious blood disease, Rules based expert systems
- More History…
- Summer of 1956 researchers got together at Dartmouth to start laying foundation in researching and developing artificial intelligent human being.
- Design goals of artificial intelligence
- Knowledge Representation – John McCarty invented Lisp, language to helpd define knowledge
- Planning (including navigation)
- Natural language processing
- Perception. How doe we feel, hear, smell, things in the world
- Generalized Intelligence. (including emotional intelligence, creativity, moral reasoning, intuition, etc.)
- 1960’s – Translation – 60 Russian sentences in to English. 701 translator – punch card demo – no semantics…context idioms…meaning was lost.. Funding killed for a decade
- 1970’s – Micro-worlds understanding language in smaller contexts. Eliza – talk therapy…
- Pick up a blue block
- put blue bock on top of red block…
- 1980’s – expert systems –
- Deep learning. McCullough and Pitts, 1940’s modeling computer on the way the brain works…
- Effectively let’s write code that mirrors the neuron to neuron system in the brain , using weights
- 2012 – Andrew Ing used 10m YouTube videos Google brought data and scale… It found cats, they recognized 16% of other objects in the stills. They did not have to tell it what to look for. it found it. THis is a Data Up approach
- why7 now…
- How does it work
- Define the number of layers, this is the deep… then give each layers some neurons… links are build between the neurons in each layers strong/week… the system will then determine what’s in the data… setting the weights on each connection,
- Artificial intelligence -> Machine Learning – >deep learning
- AI can make humans better…
- 15 to 25 years to reach general AI… has been the same for the last twenty years.
- We do not understand enough about ourselves…
- Collective intelligence will most likely lead the way. Human/human computer/human, computer/computer
- The theory of multiple intelligences was developed in 1983 by Dr. Howard Gardner, professor of education at Harvard University. It suggests that the traditional notion of intelligence, based on I.Q. testing, is far too limited. Instead, Dr. Gardner proposes eight different intelligences to account for a broader range of human potential in children and adults. These intelligences are:
- Linguistic intelligence (“word smart”)
- Logical-mathematical intelligence (“number/reasoning smart”)
- Spatial intelligence (“picture smart”)
- Bodily-Kinesthetic intelligence (“body smart”)
- Musical intelligence (“music smart”)
- Interpersonal intelligence (“people smart”)
- Intrapersonal intelligence (“self smart”)
- Naturalist intelligence (“nature smart”)
Unit 3: Combining people and computers [± 1 hour 20 minutes]
- All real uses of AI involve people and computers, people are:
- Creative software
- selecting applications
- fixing problerms
- actions only a human can do
- We are not just trying to optimize computer systems, they are human-computer systems.
- What tasks should computer do, vs people?
- machines can remember hug amounts of info
- people interact flexible with other people
- We should not think about how computers will replace people, but how we can do things better together, and new things together.
- Think Google man made content, indexed and cataloged by Google’s algorithms… Lost some reference librarians, but gained many jobs in search and advertising. and Wikipedia bots checking content as it’s actively updated… CSAIL cyber security systems machine can find event, people can reason the event… find 3x more events..
- Four roles computers can play:
- People and computer working together: combination was more accurate and more robust.
- Crowdforge – machine and people working together, with the machine having oversight and coordination
- Cogito… listening to customer calls to assess tone and mood, to inform rep.
- How can this system improve over time?
- Ever evolving systems, learning from experince top get better and better over time.
- Cyber human learning loop
- Humans get better
- programmers improve machines
- Machines learn from experience, with various forms of machine learning
- Strategic organization need to: think carefully about how we divide the tasks betwen humans and computers, and constatnly learn from experience.
Unit 4: How to gain strategic advantage [± 2 hours 4 minutes]
- Professor Malone begins by describing Michael Porter’s framework for understanding how organizations can gain an advantage over their competitors. Porter identified three generic strategies that companies can use:
- Cost leadership (being the low-cost producer)
- improving operations, loss prevention, using robots,
- Differentiation (being unique on dimensions that customers value, such as quality)
- incorporating new features that could not be identified before. Look at Google assistant Watson and patient data.
- Focus (tailoring products to a narrow segment of customers)
- Uniques need of customers and individuals. Think recommendation from Netflix and Amazon.
- In general, your strategy for using AI should be consistent with your organization‘s overall strategic approach. However, in some cases, AI may make an entirely new strategy feasible for your organization.
15 key moments in the story of artificial intelligence.pdf2017-report.pdfMIT AI M1 U1 Video Transcript.pdfMIT AI M1 U2 Casebook Video 1 Transcript.pdfMIT AI M1 U2 Casebook Video 2 Transcript.pdfMIT AI M1 U2 Casebook Video 3 Transcript.pdfMIT AI M1 U2 Casebook.pdfMIT AI M1 U3 Video 1 Transcript.pdfMIT AI M1 U3 Video 2 Transcript.pdfMIT AI M1 U4 Video Transcript.pdfMIT AI _ M1U1 Tom Malone.mp4Preparing for the Future of Artificial Intelligence.pdfWhat Managers Need to Know About Artificial Intelligence.pdfWhat can AI do right now.pdfai100report10032016fnl_singles.pdf
- Asymmetric Motivation
This is a term that describes when an incumbent gladly gives up a lower-end part of the market because the company is motivated to move up market. The motivation to flee is most often fueled by higher gross margins. This is often called “flying under the radar”. For more on this term, read here. http://hbswk.hbs.edu/item/the-innovators-battle-plan.
- Autonomous Unit
An organization that for all intents and purposes operates as a separate company with different resources, processes and profit formula(we use “profit formula”) than the parent company. This includes support functions (HR, finance, operations, marketing) and most importantly sales teams.
- Bad Money
Whether money is “good” or “bad” depends upon the circumstances in which a business finds itself. Emergent Strategy: conversely to good money, bad money is impatient for growth and patient for profit. This type of money will likely get you into a self-reinforcing spiral of inadequate growth because you haven’t tested the profitability of the market.
Deliberate Strategy: conversely to good money, bad money here is patient for growth, but impatient for profit. In this case, the laser focus on target margins will prevent the business from achieving its growth targets.
- Business Unit
An organization or organizational subset that is independent with regard to one or more accounting or operational functions.
- Celeron Processor
A product that Intel Corporation first introduced in 1998 as the processor at its very low-end of the product line. It has lower performance, and lower margins, than the rest of Intel’s product portfolio, yet has achieved the purpose of defending the low-end market from AMD.
The process by which goods that have economic value and are distinguishable in terms of attributes (uniqueness or brand) end up becoming simple commodities in the eyes of the market and/or consumers.
- Compensating Behaviors
Behaviors people exhibit because no existing solution adequately solves their problems. Customers stretch a product to do something it was not designed for, or “hack” together several products to produce a less than optimal solution. (see Scott Anthony’s book, The Innovator’s Guide to Growth)
An Idea of theory containing various conceptual elements, typically considered to be subjective and not based on empirical evidence.
- Core Competence
The main strengths or strategic advantages of a business. A combination of pooled knowledge and technical capacities that allow a business to be competitive in the marketplace. These tend to be difficult for competitors to replicate.
- Correlation vs Causality
Correlation: the extent to which two datasets are related to each other. Statistics such as “Millennials are 2.5x more likely to make New Year’s resolutions” rely on a high correlation factor between the dataset “Millennials” and “New Year’s resolutions”. Being a millennial doesn’t cause you to make a new year’s resolution.
Causation: the extent to which one dataset or event causes another. Understanding what causes what and why is the focus of the entire physical science world, yet in business we often rely heavily on correlation to understand our customers (to our own detriment).
- Cost of Goods Sold (COGS)
Costs include all costs of purchase, costs of conversion and other costs incurred in bringing the inventories to their present location and condition. Costs of goods made by the business include material, labor, and allocated overhead.
The processes and values embedded into an organization by assumption, rather than by conscious decision.
- Customer Segmentation
This is the practice of dividing a customer base into groups of individuals that are similar in specific ways relevant to marketing. Using the lens of disruptive strategy, it is critical to segment customers by circumstances rather than by demographics.
- De-coupling Point (Interdependence vs Modularity)
The hand-off point between subsystems (products) or customers and suppliers (supply chain). As a product/service/industry becomes more modularized, many de-coupling points arise as the interfaces between subsystems or the supply chain become well defined. If a product/service/industry has an interdependent architecture, and subsequently a business adopts an integrated strategy, de-coupling points will be few and far between.
- Deliberate Strategy
A deliberate strategy is one that arises from conscious and thoughtful organized action. It’s generated from rigorous analysis of data on market growth, segment size, customer needs, competitors’ strengths and weaknesses, and technology trajectories.
Shifting out of parts of the value/supply chain to focus on a specialized strategy. Companies should disintegrate in a world with a modular architecture.
The classic definition of disruption is a disturbance or problem that interrupts an event, activity, or process. With a traditionally negative connotation, this term has taken on a life of its own in the business world to basically mean anything that is counter to the norm. As it pertains to D-Strat, we will define it more precisely.
- Disruptive Technology
- Efficiency Phase
When companies sell mature products or services to the same customers at lower prices. Companies accomplish this by developing a business model that can still make money at lower prices-per-unit sold in order to increase profitability.
- Emergent Strategy
An emergent strategy is one that arises from unplanned actions from initiatives that bubble up from within the organization. It is the product of spontaneous innovation and day-to-day prioritization and investment decisions made by middle managers, engineers, salespeople and financial staff.
- Getting the Categories Right
It is part of human nature to put things into categories; it helps us make sense of the universe in which we live. The concept of “getting the categories right” with regards to disruptive strategy is understanding the circumstances underlying a phenomenon, rather than surface-level attributes, and creating categories based on these. For example, if you are attempting to predict adoption rates of a new in-home IOT device, it would be unwise to use the adoption rates of the iPhone as a comparison because “hardware” is an attribute-based category and not a circumstance-based category.
- Good Money
Whether money is “good” or “bad” depends upon the circumstances in which a business finds itself. Emergent Strategy: during the nascent years of business, good money is patient for growth but impatient for profit. Money needs to be impatient for profit to accelerate a disruptive venture’s initial emergent strategy process. It forces management to test as quickly as possible the assumption that customers will pay a profitable price for a product.
Deliberate Strategy: upon switching to a deliberate strategy (after exiting the emergent strategy process), good money is now money that is impatient for growth, but patient for profit. This is because the profitability of the market has been fully tested and now the business needs to invest for growth.
- Gross Margin
Gross Margin = (Revenue – COGS) / Revenue
This refers to setting up check-points while executing a strategy plan to ensure you are translating strategic concepts into tangible, practical plans of action correctly.
Simply put, an incentive is a thing, typically money in business context, that motivates or encourages one to do something. The sales team incentive structure is of huge importance in determining whether your company’s intended strategy will indeed be your actual strategy.
Adding parts of the value/supply chain to deliver on an integrated strategy. Companies should integrate in a world with an interdependent architecture.
- Integrated Strategy
This strategy focuses on the entire system to deliver an end product/service to meet the minimal satisfactory performance requirements for the customer. An integrated strategy is best suited for industries with an interdependent architecture, i.e. each component in the product/service is design dependent on each other and cannot be separated.
- Interdependent Architecture
The structure of a product or service if one part cannot be created independently of the other part – if the way one is designed and made depends on the way the other is being designed and made. Optimize performance. (Innovator’s Solution, Page 127, Getting the Scope of the Business Right)
The place where any two components fit together. Exists within a product, as well as between stages in the value-added chain, i.e. Design -> Manufacturing -> Distribution.
- Job to be Done
A job-to-be-done is defined as the progress an entity, customer or business, is trying to make during the course of day-to-day life. A job-to-be-done is a circumstances-based description of understanding your customers’ desires, competitive set, anxieties, habits and timeline of purchase.
Integrating around the job-to-be-done is how a company organizes itself and product/services offerings to deliver on a set of experiences that perfectly “nail” the JTBD. This means instead of organizing around traditional categories, i.e. marketing, product development, sales, etc., companies should organize and integrate in order to deliver a product/service perfectly centered around the JTBD.
- Low-End Disruption
A product, service or business model that enters the market at a lower performance and price point than an existing offering. A low-end disruptor has the following characteristics:
1) Initially, target “overserved” customers who are unattractive to incumbents due to low margins
2) Product/service performance is simple, yet deemed “good enough” compared to existing solution
3) Incumbents either ignore or cede the market easily to the new entrant. This is called asymmetric motivation.
- Market Creating Phase
The early stages of a new product or service when a company is focused on the development of the product or service to meet the customer’s job-to-be-done.
- Modular Architecture
The structure of a product or service where the fit and function of all elements are so completely defined that it doesn’t matter who makes the components or subsystems, as long as they meet the specifications. Optimize flexibility. (Innovator’s Solution, Page 128, Getting the Scope of the Business Right)
- Net Margin
Net Margin = (Revenue – COGS – OpEx – Interest & Taxes) / Revenue
- New-Market Disruption
A product, service or business model that creates an entirely new market by targeting non-consumption and offering inferior performance according to traditional metrics, but superior performance according to new metrics. These new metrics often focus on simplicity and convenience. A new-market disruptor typically has the following characteristics:
1) Targets traditional non-consumption by focusing on the underlying Job-To-Be-Done, Since we don’t give examples for the other ones I think we should keep it consistent. If we want to keep this example then I think we should add examples to the others. But as this is a glossary I don’t think it’s necessary.
2) Incumbents don’t see the entrant as competition because of different performance metrics and/or product/service characteristics.
3) As the new-market disruptor gains market share, incumbents are unable to respond, even if they want to, because they cannot compete on the new performance metrics.
This term describes how we traditionally think about people who aren’t buying a product in a category.
To compete against non-consumption, introduce a product/service that is focused on the Job-To-Be-Done.
- Performance and “Good Enough”
Performance: the criteria your customers determine to be important in product use and adoption. Performance is usually defined by metrics relating to the end-user experience, but also may be defined by inherent product metrics as a proxy for the end-user. Understanding how your customers define performance is critical to success and understanding your true competitors.
Good-Enough Performance: this is the performance point at which your customer adopts your product. Understanding what performance is “good enough” enables you to know whether to adopt an integrated or specialized strategy for your industry, as well as understand potential entrants.
- Performance Defining Component
The component in the value stack that provides the functionality that customers care most about. This is typically where the most profit can be made. It is IMPERATIVE to understand this can change. For example, the performance-defining component for Intel Microprocessors in the early 2000s was speed, or cycles per second (hertz). With the advent of Wifi and the laptop form-factor, the customer’s performance defining component shifted to battery life. This shift caught Intel flat-footed, even though Intel was the one who developed Wifi in the first place, and they lost a substantial amount of market share to rival AMD. Intel eventually was able to weather the storm and shift its product line to this new performance-defining component, but not without a lot of organizational and financial pain.
- Permission to Grow
Any business needs the implicit permission of society, as manifested in laws, regulations and taxes, to grow, operate and simply exist. In the new sharing economy, laws/regulations have changed to grant or refuse permission for this new business model. See this article by HBS Professor Derek van Bever on Uber and the permission to grow. https://hbr.org/2015/02/uber-needs-our-permission-to-grow
The patterns of interaction, coordination, communication and decision making through which the transformation of resources into products are accomplished. They include the ways that products are developed and made and the methods by which procurement, market research, budgeting, employee development and compensation, and resource allocation are accomplished. (Innovator’s Solution Page 183)
- Product Architecture
How a product’s components and systems interact – fit and work together – to achieve the targeted functionality.
- Profitability vs. Profit Formula (Charles Schwab Case)
Profitability: measured with income and expenses, using both real numbers as well as ratios, i.e. net profit, net income, net margin, etc.
Profit Formula: the profit formula is how organizations internally determine which projects to select. Often consists of specific ratios targets, i.e. X% gross margin, IRR of Y, etc.
- Purpose Brand
A purpose brand links customers’ awareness that they need to do a job with products that have been designed to do it well. The highest level in the job architecture.
- Resource Allocation Process
The process by which resources are deployed to drive initiatives within a business. There is always a process, regardless of whether it is explicitly stated or otherwise. A resource allocation process alone isn’t “good” or “bad”; it’s just the process. The question should be, does the process prioritize the initiatives we’ve strategically said we want to take on? Does it create the right incentives that align with our strategy?
People, equipment, technology, product designs, brands, information, cash, and relationships with suppliers, distributors, and customers. Usually people or things – they can be hired and fired, bought and sold, depreciated or built. (Innovator’s solution, page 178)
- Return on Net Assets (RONA)
RONA = Net Income / (Fixed Assets + Working Capital), where working capital = current assets – current liabilities
- Skate to where the "money/puck" is
A phrase popularized by Wayne Gretzky, a famous Ice Hockey player, when discussing the success of his career. The analogy to business is that instead of focusing on where the money is today, use business theory to understand where the money will be tomorrow and orient your company towards that future.
- Specialized Strategy
This strategy focuses on one piece of a system and doing that piece superbly. A specialized strategy is best suited for industries with a modular architecture, i.e. the inputs/outputs and clearly defined and standardized.
- Sustaining Innovation
A product, service or business model that provides performance improvements in attributes most valued by the industry’s most demanding customers. The improvements may be incremental or breakthrough. A sustaining innovation typically has the following characteristics:
1) For new entrants, the incumbent has every motivation to fight your entrance to the market and sees you as a direct threat.
2) A product/service improves upon traditional performance metrics and charges more money for the improved performance.
3) For incumbents, uses the existing company’s resources, processes and profit formula to develop and execute the offering.
- Sustaining Phase
After the product or service has been defined and deployed into the marketplace, the company tries to evolve the product or service to meet the needs of the best customers in the market in order to beat the competition.
- The Innovator's Dilemma
In The Innovator's Dilemma Professor Christensen introduces the concept of disruptive innovation.
- The Innovator's Solution
In The Innovator's Solution Professor Christensen offers frameworks to help business leaders face disruption. These ideas may be applied to companies wanting to become a disruptor or to incumbents who are trying to avoid disruption.
- The Role of the CEO
The CEO, or other similarly high ranking executive, is critical at the beginning of the creation of a disruptive growth engine. Because the processes and values of the mainstream business by their very nature are geared to manage sustaining innovation, there is no alternative at the outset to the CEO or someone with comparable power assuming oversight responsibility for disruptive growth.
- Value Stack
The structure of how value is created and captured in a product/service/industry.
The standards by which employees make prioritization decisions – those by which they judge whether an order is attractive or unattractive, whether a particular customer is more important or less important than another, whether an idea for a new product is attractive or marginal, etc. (Innovator’s Solution page 185). Values and prioritization decisions in for-profit businesses are almost always dictated by the profit formula.
- Two sources of Strategy
- Strategy is process, not an event. Understanding the process is a key to success. Understand the difference between "good money" and "bad money."
- By the end of this module you will be able to:
- Understand the strategy development process within your own company
- More effectively manage the strategy development process
- Determine which "type" of money will be optimal to grow your business
- Resource allocation: Everyday the team has to decide what to focus on, and what not to focus on. We need to look at what the company needs to prioritize financially, then determines what get’s implemented and what does not, this becomes the strategy. The allocation of resources is not an event, followed by implementation. It is actively happening.
- The resource allocation process will determine which deliberate and emergent initiatives get funded and implemented, and which are denied resources.
- If you want to see the real strategy of a company. Don’t listen to what they say, watch what they do.
- Below is the entire resource allocation process. A useful exercise would be to diagram this out for your own organization
- As companies grow, it makes it hard to see small opportunities. DO not become a monolithic company. Have small business units that add up to a large company, then small opportunities look bigger to the smaller business units.
- The resource allocation process will automatically prioritize initiatives based on the profit formula for the company, whether good or bad.
- Intel had a deliberate strategy that was to be the leaders in smart phone processors, however the profit formula/resource allocation process just wouldn’t let it happen.
- Need to set a new criteria (profit formula) for the resource allocation process.
- Profit formula controls the resource allocation process.
- process can prevent innovation. Thus we need a new business model to go after the new business.
- good, emergent ideas often come from lower-level employees. Thus, implementing processes to surface these ideas is critical.
- The HR process that promotes based on individual $$ contribution, is based on an ideal that will reward showing value every 18 to 24 months. If an innovation will take longer than 18 to 24 months, they will will probably be passed up.
- There are three phases of business growth. Here are a few bullet points to crystalize these stages in your mind:
- Market Creating Phase (job creation) – A deliberate strategy, but know that innovation will come about via an Emergent Strategy, as well.
- The early stages of a new product or service when a company is focused on the development of the product or service to meet the customer’s job to be done.
- Sustaining Phase (no net new job growth) – Deliberate strategy, top down approach to grow the business.
- After the product or service has been defined and deployed into the market place, the company tries to evolve the product or service to meet the needs of the best customers in the market in order to beat the competition.
- Efficiency Phase (job loss)
- When companies sell mature products or services to the same customers at lower prices. Companies accomplish this by developing a business model that can still make money at lower prices-per-unit sold in order to increase profitability.
- IBM/Sears introduced Prodigy
- Invested $500m each ($1b)
- started with a focus on shopping.
- saw people emailing, and decided to charge them.
- Instead of seeing email as an emergent opportunity, leaders at Prodigy focused on their deliberate strategy to facilitate an online portal where users could access information and shop.
- AOL focused on email, and a purpose brand "You’ve got mail"
- It’s hard to get a customer to "change the job", but if you help them do their job , but better, you win.
- You can;t beat the resource allocation process, you need to manage it.
- Lesson from AOL
- Never believe that the strategy that helped us to be successful, will not always be the strategy that keeps us successful.
- Change in the strategy is not an event, it’s a process
- Continue to pursue the existing strategy, but build a new BU to be the next business.
- Blockbuster: Acquire movies for all 5000 locations, rent as many as possible in 3 wks, then clear. This required late fees as a tax for not being able to rent again.
- Netflix went after the DVD market, at a flat rate.
- as DVD became more and more available. they switch to rental and late fees.
- then moved to unlimited rentals for a flat fee.
- Implemented a recommendation system, this spread out demand. The movie had to be in stock to be recommended.
- Blockbuster tried to add an online business, but the resources, processes, and profit formula were deployed differently not allow for them to work together long term. The business was built to support stores.
- Blockbuster merge what was the best in their traditional business with their online business.
- Netflix created a separate business around streaming.
- Now they continue to move up to content.
- "Plans are useless, but planning is indispensable." Dwight D. Eisenhower
- Good Money and Bad Money
- The basic idea of good money and bad money is that the type of money a manager accepts carries specific expectations that must be met. These expectations heavily influence the types of markets and channels that a venture can and cannot target. The very process of securing funding forces many potentially disruptive ideas to get shaped instead as sustaining innovations that target large and obvious markets. Thus, the funding received can send great ideas on a march towards failure.As emergent ideas are being nurtured during nascent years, money must be patient for growth but impatient for profits.When winning strategies become clear and deliberate ideas need to be carried out then money should be impatient for growth but patient for profit.
- Company good money, bad money lifecycle.
- Successful companies
- company faces growth gap (shareholder expectations must be exceeded)
- Good money becomes impatient for growth
- executive temporarily tolerate losses
- mounting losses precipitate retrenchment.
- End up back where you started, with a growth gap.
- Again, focus on the "Jobs to be done." If you focus on this, then the chances of good money and bad money tripping you up is minimized.
- Project Beacon, leveraging car deployments, Hughes electronics, and EDS
- Very emergent strategy
- Approaching the market with a broad strategy is actually consistent with the emergent strategy process. During the market creating phase, OnStar didn’t know what services would resonate most with customer jobs to be done. Chet was effective at allowing new ideas to surface and be tested in the marketplace. However, as Chet pointed out, the "swiss army knife" approach isn’t viable long-term. Chet needed to find the winning strategy that would allow OnStar to rally around a deliberate strategy.
- safety and security emerged as the key value proposition.
- Gen 2 hardware was 1/3 the cost of Gen 1.
- GM was focused on RONA (return on net assets) as the profit formula
- OnStar settled on the broad job to be done of "help me have safety, security, and peace of mind"
- "The purpose brand gives you license."
- The company (GM) was patient with results.
- Active engagement of most senior people in the company. Would have been great to have split RPP (Resource, Processes, and profit formula.)
- Managing Your Strategy Development Process
- Senior management must simultaneously, yet separately manage the strategy development process.
- Leaders must seek Good Money based on your current situation.
By the end of this module, you will be able to:
Build a stronger intuition for where the profitability will be in your industry
Determine which operations are critical to be done in-house and which operations can be out-sourced
Better predict the direction of your industry and company
Things managers get wrong:
do core competence inside
let other stuff happen outside
“skate to where the puck is going, not to where it is.” gretsky
Skate to where the money will be
Interdependence and Modularity
Interdependence: Design of one piece, depends on the design of the other pieces (unpredictable interdependencies)
Modularity: How pieces fit together a re well defined, and theoretically easier to modify. (no unpredictable interdependencies)
Modular architectures, lend to disintegration
In the early stages of any industry, products and services tend to be very interdependent. Over time, as products and services become more defined, these same products and services become modular.
as you mature you have to become more modular to become faster, flexible, and responsive.
What is the right level of integration
If you have a superior product, that is not modular to just “plug and play,” then you must integrate backward/forward into the product until you reach modularity. this can be costly.
DuPont found that the interface between kevlar and tires was unpredictable and not-good-enough. To be successful in the tire industry, DuPont would have had to integrate forward to become a tire designer. Only then could DuPont manage all of the interdependencies between kevlar and tires. DuPont’s only other option was to wait and hope that an external tire designer would figure out the interdependency.
they found applications where they could simply pull out the old, and plug in the new.
RCA had color TV, but no color content.. RCA and NBC merged
Three types of interdependence:
Functional or Technological Interdependence
The interdependence between two components of a product or service.
Example: The way kevlar interacts with a tire is unknown and thus, interdependent.
Profit Formula Interdependence
The interdependence between a product or service and the way a company makes its money.
Example: Organizations that make fighter pilot jets are forced to serve the high-end of the market where they can earn larger margins to cover their high fixed costs. Because of this interdependency, high-end jet organizations can’t target low-end drones because their high-fixed costs don’t allow it.
Marketing or Brand Interdependence
The interdependence between a product or service and how it is marketed.
Example: Consumers didn’t connect with “kevlar belted radials” because the word “kevlar” didn’t mean anything to them. However, consumers embraced “steel belted radials” because the word “steel” connotes strength and durability. (Note: The interface between kevlar and tires is an example of both functional and brand interdependence.)
Knowing when to disintegrate/become more specialized
IBM: heads&disks>drives>computers>software>services (decoupling point)
> data > cognitive?
If you do not integrate forward to the decoupling point, then you have to sit and wait for someone else to come along. This typically does not happen.
Once the industry became “good enough” IBM needed to disintegrate; this was very costly and timely, and IBM’s profitability took a nose dive, and they got out of the less profitable “heads&disks>drives>computers” businesses.
It is interesting how your greatest strength can become your greatest weakness if you fail to switch your strategy from integrated to specialized.
Avoiding Commoditization & Finding the Performance-Defining Component
As products and services become good enough, their architecture becomes more modular. This allows companies with a specialized strategy to do one piece of the system very well. Over time, as a growing number of new competitors enter the market, commoditization occurs.Essentially, what happens is that barriers to entry are lowered as the various interfaces within the system become more defined (or more modular).
As commoditization develops within an industry, profits decline. Companies are forced to find profitability in new ways. In other words, companies must seek the performance-defining component.
The performance-defining component (or subsystem) is defined as the component in the value stack that provides the functionality that customers care most about. This is typically where the most profit can be made.The performance-defining component itself tends to be not yet good enough for customer needs. As a result, it is usually more interdependent than modular. Thus, the barriers to entry are high and attractive profits are available for few competitors.
Need to go to where the money is. this will shift over time. We are successful cause we are “here” at the right time.
Continually “skating” to the performance-defining component is critical to remain viable long-term. Another principle that will help you stay focused on the performance-defining component is the principle of the job to be done. If you can maintain a discipline to build products and services around the job to be done, your chance of continuing to produce the performance-defining component as commoditization occurs is greatly increased.
Integrate company around job to be done. (ruggedness?) If you do this differentiation is natural.
what we need to integrate and how we integrate to deliver an experience is hard for an attacking company to surmount.
est. 1997, based in Taiwan
chips, chips in optical disc reader, chipsets in mobile, on chip solutions
MediaTek integrated forward to the decoupling point. Turnkey solution
Low entry barrier
faster time to market (3 months to six months to market.)
allowed customer to add differentiation to the phone.
Leveraged “reference design”, could tweak, qualified components for verified vendor list
What if you were able to design a phone per a customers specific job to be done; service and support them all the same.
$200m to $7B
More business model disruption vs technology disruption.
Discovering the scope for your organization
Here are some takeaways to keep in mind as you try to maintain the disruptive scope of your organization:
Always consider your strategy as being “temporarily successful.”
“Skate” to the “performance-defining component.”
Defend your competitive position by organizing around a “job to be done.”
Frequently refer back to “these theories to guide your strategy like a compass.”
- Questions we will explore:
- Individuals: Are the people good enough to succeed?
- Organization: Will our organization have the ability to succeed at this task?
- Goals of this module, to understand:
- What can the organization do, and what it can not do.
- Identify the resources, processes, and profit formula required to succeed.
- Build the capabilities you need to grow.
- Three factors (RPP)
- Resources (tangibles, visible, can be hired and fired, bought and sold, depreciated or built – flexible and transferable – Resources are the things that are available.)
- Technology, People, Products, Facilities, Equipment, Brands, Information, Distributors.
- Processes (how resources work together to get the outcomes you need. To understand what a company can and can not do you need to understand processes. decision making, communication – inflexible; not meant to change – Processes tells you how to do it.)
- Product development, Procurement, Market research, Budgeting, Employee development.
- Process creation
- Task emerges and people need to work together to get it done.
- If fails, they re imagine the task
- Try task 1.1 execution again, if it passes, use it again… This creates a processes.
- Task emerges and people need to work together to get it done.
- Processes that people don’t even think about it how to do it, they just do it, becomes the culture.
- When people follow a process to do a task for which it was designed, it usually works efficiently. But when the same efficient process is employed to tackle a different task, it often doesn’t work. In other words a process that becomes a capability in executing a certain task can be a disability in executing other tasks.
- Profit Formula (any person in the company who is going to prioritize this over that. e.g. today I’m going to call customer A, instead of customer B…. based on profit formula. Very diffused, decentralized.. Criteria is critical, if asked to do something is not a lined with the cireteria you can see that the company can not do it. – Criteria in Profit formula tells you what a company can not do.)
- Gross margin targets, ROI/ROA, thresholds, utilization goals, type pf orders or customers.
- What your organization cannot do
- Early in a company, all capabilities exists in the resources.
- When a company grows, the ability shifts from resources to to processes.
- Do we have the resources to succeed, will processes enable us, can we prioritize?
- Find the ability to do something is rooted, will tell you what a company can and can not do.
- Good people, working in an organization that is not capable, are destined to fail.
- NYPRO Healthcare
- Management that has been around for 30+ years
- Founded in 1955 Frank Kirk and Gordan Stoddard (50% to 100%)
- Customers want “Lowest cost per cycle.” They pushed “high cavitation” within a tool, so they are able to extrude more parts in less time.
- Gordan was not a formally trained engineer. His thing was understanding people and winning them over.
- netstal machines were used.
- Technology: Nypro invested in and developed top-of-the-line equipment in injection-molded plastic manufacturing
- CEO Gordon Lankton, who had an innovative mind-set and attracted great talent
- Strong general managers of each plant
- World-class engineers
- Facilities: Global plant network that allowed Nypro to be close to customers
- Brand: Recognized name for high-quality injection-molded plastics
- Benchmarked plants against each other via a daily and weekly reports to spot areas for improvement (e.g. tool turnover time)
- Strong base of central talent, but allowed innovation to come in through different plants
- GM: Good to go out and see what others are doing, to “steal” and make your plant better.
- Gordon’s weekly visits and person-to-person conversations to find ways to improve
- Yearly General Manager meeting to gather and share best practices
- Profit formula:
- Fewer, large customer accounts or small accounts with the potential to grow
- High-volume, long-run orders that maximize machine utilization
- Complex, technologically challenging orders with better margins.
- Each machine was like a hotel room. If you don’t use it one night, you lost that revenue forever.
- Sales team was off volume, engineering was off of profitability.
- Here are some of Nypro’s strongest capabilities and disabilities (everything they did was geared towards the high-volume market):
- Nypro organization can:
- Deliver industry-leading efficient manufacturing
- Complete high-volume programs with very sophisticated molding machines
- Solve challenging manufacturing problems
- Nypro organization can’t:
- Deliver simple manufacturing with quick change-overs
- Complete low-volume, short runs
- Conduct fast prototyping
- Nypro Dilemma
- Gordan saw new operation in Japan
- Operation in japan, built around small machines… 60% were running.
- The machine caught his attention. it was a simple machine
- Gordan thought there was a operunityu to gain new customers small run, contaminates, grindings and oils.
- Easy switch over.
- faster cycles
- Less parts per hour
- They created a novaplast (short run, simple products). After they decided to adopt the NovaPlast machine, Nypro’s leaders had three options:
- Nypro initial decision: Integrate 1-2 NovaPlast machines into several different plants
- Concentrate all NovaPlast machines into 1-2 existing plants
- Second choice: Build a new plant dedicated to the NovaPlast machines
- They had problems with staff. they wanted to make it more complicated.
- Sales got paid on volume, so it went against existing model.
- Needed to create a new sales group for NovaPlast machines
- Seems an odd way to impart a new tech. Not sure I saw the pain they were solving
- Ultimately NovaPlast was killed
- Charles Schwab and introducing an on-line trading platform eSchwab
- Clay went from Merrill Lynch to Charles Schwab, because the latter cared for the type of client clay was.
- Schwab was a low-end disruptor
- Schwab decided to disrupt itself charging $79/trade for traditional, and $29/trade for on-line. They set the latter up as a separate business unit.
- EMC – Merging competing products through an acquisition.
- Core: information storage. 55%
- Mid-tier (SMB/SOHO) 6%
- Acquired data general clariion for mid-tier, two server model.
- data general 20% direct, 80% channel, EMC 80% direct, 20% channel
- They were going to move data general to EMC model… oh, oh.
- Formed a separate sales force for Clariion 6% to 35%, with dell as channel partner.
- Went from $400m to $4b product line.
- VMWare was treated completely differently. $60m to $6b
- Big company can kill it, or hug it so tightly it takes the breath away.
- Focus on the people in acquisitions. Let the small company use the big companies resources when needed, but do not let the large company just use the little company.
- What teaching like Clays are…Guide posts, warning posts, a way of thinking. Remember that values are non-negotiables, but culture should be flexible.
- You can not disrupt yourself
- You need to set up a different business unit, under the umbrella of the existing business
- If sustaining then fold the innovation/conpany in.
- if disruptive, allow it to operate separate.. If you fold it in, you will ultimately kill the innovation.
- When companies succeed, sometime they need to develop new capabilities.. They need to be built independently of the old.
- This can help us to predict when we need a new business unit, and when we need to craete a new one.
- If a manager see that they need new resources, proceesses, and profit formula… then the game will be over.
- You need to be looking out into the future.
- We need to think deeply to understand what an organization can and can not do. It tells use where we need to build abilites. telling use where to create new, of leverage old
- resources – flexible
- processes – inflexible
- profit model – dictates criteria to prioritize “this” over “that.”
- We need to think deeply to understand what an organization can and can not do. It tells use where we need to build abilites. telling use where to create new, of leverage old
- Jobs to be Done:
- Understand the job the customer is trying to do, and the develop a product that nails this job perfectly. The probability that your innovation will be successful will improve in dramatic ways.
- Marketing: Casual reason why customers buy your product.
- 75 – 85% of new products fail, and do not succeed financially. because they do not target a job
- Companies focus on correlation vs causation.
- The goals for this module are:
- To see that many products fail because they are focused on correlation, not causality
- To find important “jobs” in the market for which your products might be “hired”
- To understand your products’ true competitors from the customer’s perspective
- To figure out what and how you must integrate in order to get a “job” done perfectly
- If you do this right, it makes it harder for others to disrupt you.
- a major key to discovering customer jobs to be done is to observe and interview consumers.
- customer segmentation:
- I have a job to done, and I look around and pull it in to my life.
- Include substitutes in the size of the market.
- Focusing on the job to be done helps you answer some critical questions:
- What “job” does the customer hire your product to do?
- How big is the market, and is it growing?
- Who is competing to help the customer do this “job”?
- A problem or an opportunity that someone is trying to solve. Job = needs to be done. We hire products are services to solve the problem.
- Help me…
- I need to…
- Help me avoid…
- Not job to be done (more a part of experience need to provide to customer gets the job perfectly)
- Low cost
- Once you know the job, you can clearly see who the true competitors are, how large the market is, what opportunities there are for innovation, and how to market the product.
- If you frame the business in products, you will come and go. If you frame it around the job, it frees you to change the way you deliver on that.
- Two dimensions of “Job to be done”
- Functional: Tangible and measurable
- Emotional and social: How it makes me feel
- Disney does the “Job to be done” well.
- Why Disney: Fun with family and friends, Escape reality, Kid again, Meet characters, Rides & attractions
- Jobs to be done stay focused over time. Were a focus on products can leave thing unpredictable.
- When you figure out a better way to get the job done. It always results in financial and market success.
- Job architecture 4 layers
- Purpose Brand: Word that pops in to customers mind, when they have a job to do. Mind share.
- Integration: What needs to integrate to provide the experiences
- Experiences: What are the purchase and use experiences that will sum up to nailing the job perfectly.
- Job to be done: understand of what the job is to be done. Functionally and emotionally
- You can identify jobs to be done by observing:
- Yourself: Why do you do what you do?
- Current customers: Why do they buy your product?
- Non-customers: Why do they not buy your product?
- Former customers: Why do they no longer buy your product?
- Compensating behaviors: Inconvenient workarounds people use because there’s no product that fulfills their job to be done well
- Entertain, Inform, Educate in a utilitarian way… Always there.
- Not about a demographic, it’s that we have a job to do….
- How do you find jobs to be done?
- Think in geographic terms (e.g. someone out there has a job to be done. Why would they hire us)
- Reflect deeply on personal experiences
- Observe current customers (why are they buying our product, and when they do not what are they using to get the job done?) – When a customer leaves, you must find out why…
- Identify compensating factors (are people having to do work arounds)
- Why are they buying products? if they dont, what are they doing to get the job done.
- Minute clinic
- Yes/No decision
- No wait for unpredictable wait of time. Every 5 minute response.
- Job to be done: “help me to quickly and conveniently, get the healthcare I need without seeing the doctor.”
- The Godrej Group – understanding refrigeration needs in India.
- 1897 making locks – safe, furniture, refrigerators, etc, etc
- 80% of indias population does not have refrigerators.
- There is no job that is created. They are doing the job.
- Technology exist, job t be done exist, but they have not been brought together in an economic way.
- Data is not real. It’s a proxy. Data has it’s value, but you need to dive in. Get outside.
- Chotukool is new-market disruption, that is competing for non-consumption.
- Bose – QC20 Noise Cancelling Headphones.
- Job to be done: Help me to experience air travel in a more tranquil way.
- Experiences: Product tests that simulate airline travel and other annoying noises. Flexible return policy, and superior customer service
- Integrations: Placement of stores, where I shop. Advertising with NFL, engineering bent.
- Name: Bose <Tagline>
- Two views of God: 1) on top of the universe and with a stroke of the finger and effect change, or 2) inside the universe with a uncanny understand of all of it’s levers and because of this knows what’s possible and how to effect change.
- Power comes from understanding of constraints, forces, vectors, and processes. If understood perfectly, you can harness those things.
- When someone comes in the bottom of the market and moves up. “It’s not what to think, it’s how to think.”
- Align with Amazon
- Optimize supply chain around you Commerce
- Hire a bunch of technology and data scientist become a technology company
- Spend a bunch of money and stores and try to create some new special experience
- Leverage the vast real-estate become a reit
- Line with other retailers and become effectively a large-scale department store
- Creative technology
- 3 Types of innovation
- Sustaining Innovation (incumbents usually always win)
- Makes good products into better products
- Target high-end of market, more profitable customers
- Brings more profit margins
- Offering ever better products, to sell at ever better margins, to your very best customers.
- Incumbents fight new entrants
- Innovation aligns with existing business model
- Low-end disruptive
- Causal mechanism is the pursuit of profit. coming in to the low-end, low margin, part of a business will typically lend itself to less competition as the incumbent is willing to let the lower-margin business go, this can continue up the “stack” until the incumbent is no more. leaving behind low-profit businesses can sometimes open the door to disruptors. Low-end disruptive innovations take advantage of markets in which existing products offer more performance than many customers want or need. Technology advancements outpace customer need, this opens opportunity for “good enough” products
- gain market share against old
- low-end disruptive innovation takes advantage of the performance surplus, which occurs when product performance overshoots what customers can use.
- Sustaining Innovation (incumbents usually always win)
- Offers “good enough” but not much more
- targets “over-served” customers
- figured out a fundamental different business model
- Incumbents flee from new entrants, moving up market in pursuit of more valuable customers. these disruptive innovation can not be adopted by incumbents.
- New-market disruptive
- Make product more accessible. Mainframe – $2m, PC $2k, Mobile $200. What will be $20, what will be $2?
- Hard for incumbents to move into new markets for the new-market disruption. Because there is no motivation from a profit formula standpoint, it’s not their business model.
- creates new markets and will not find incumbents there. then sucks customers out.
- Honda Cub entering market where there was not consumption.
- High margin dealers did not want to sell, but lawnmower companies did..
- Attacking a sustaining market did not work, but New market innovation did.
- This opened up a new market, where only HarleyDavidson still exists as the domestic player.
- Need to have an application that would make money, vs going into a market to find money.
- new-market disruptive innovation targets non-consumers because existing products are too expensive or complicated to use.
- Targets “non-consumption.” people who did not have ability or access to incumbent product
- Make profit for lower price-per-unit sold than incumbent tech. Think dollar vs percentages.
- Product provides lower performance for the existing market but higher performance for non-consumers.
- Incumbents flee from new entrants, these disruptive innovation can not be adopted by incumbents.
- New-market disruptive
- No technology is intrinsically sustaining or disruptive, that is given based on deployment/implementation.
- Sustaining or disruptive is relative to existing business model or products
- If an innovation can deploy into an existing structures (business model/products), it’s sustaining. If the innovation can not be deployed into existing structures, then it is disruptive.
- Many innovations can be considered both low-end disruptions and new-market disruptions.
- You can not disrupt yourself. Because the core company can not make money on the disruptive innovation.
- They will take technology and implement it in the way that serves the company, making it a sustaining innovation.
- In order to do it you need to set up a company with a different business model.
- If you try to move into the old business you will ruin what you just built/acquired.
- Set up a completely different business unit and let them disrupt the core business.
- Retail: 1960 316 traditional department stores, 1962 discounters entered the market. Only 8 traditional department survived. Only one successfully transitioned into a discounter, it was Dayton-Hudson, that created the separate business unit “Target.”
- Circle UP (consumer and retail funding)
- comparables for analysis
- Capital raise 8-12 months to 61 days,
- Charge less money, and
- Partnerships (distribution, product placement) and services (advisor network)
- Disruption is typically an opportunity well before it’s a threat. and if you go after low-end disruptive innovation, then you can take share. It’s only a threat when incumbents start to fight back.
- You need to start innovating now, while the core is strong. The core will classically be strong, while the disruption grows. Do not wait until Core atrophies.
- Sustaining innovation can be merged in to the core.
- Here are some guidelines to use when trying to get disruption to work in your favor:
- Disruption is typically an opportunity long before it’s a threat
- You must begin to innovate while your core business is still strong
- Allow disruptive businesses to run independently of the core business
- Spot disruption by observing customers at the bottom of the market
- Protect your business by focusing on and integrating around the job to be done (a topic introduced in another module)
“No muscles without strength, friendship without trust, opinion without consequence, change without aesthetics, age without values, life without effort, water without thirst, food without nourishment, love without sacrifice, power without fairness, facts without rigor, statistics without logic, mathematics without proof, teaching without experience, politeness without warmth, values without embodiment, degrees without erudition, militarism without fortitude, progress without civilization, friendship without investment, virtue without risk, probability without ergodicity, wealth without exposure, complication without depth, fluency without content, decision without asymmetry, science without skepticism, religion without tolerance, and, most of all: nothing without skin in the game.” ~ Nassim Nicholas Taleb
When seeking to develop an innovation strategy, here are some questions you should get answered.
How do I see emerging trends before they become problematic?
How do I generate a robust pipeline of new growth ideas to consider?
How do I identify and focus on the highest-potential opportunities in areas like blockchain and AI?
How can I motivate traditional company management to realize the need for digital transformation?
How do I evaluate competitive signals in a noisy, buzzword-filled market?
How do I get the middle layer of my company to embrace change?
How do I bring outside ideas into my organization?
When does it make sense to be a fast-follower? And when does it not?
How do I decide whether to build, buy or partner?
Should I start a venture capital fund?
Impact on developing economics and jobs
A realistic view
Goldilocks rule for AI:
Too optimistic: Sentient/AGI, killer robots
Too pessimistic: AI cannot do everything, so an AI winter is coming
as opposed to the past, AI is creating value today.
Just right: Can't do everything, but will transform industries
Limitations of AI
performance limitations. (limited data issues)
Explainability is hard (instructible)
Biased AI through biased data
Bias against women and minorities in hiring
Bias against dark skinned people
banks offering hiring interest rates to minorities
reinforcing unhealthy stereotypes
"Zero out" the bias in words
Use more inclusive data
More transparency and auditing processes
More Diverse workforce
Minor perturbation to pixels can lead and AI to have a different B output.
Defenses exist; incur some performance cost
There are some applications that will remain in an arms race.
Adverse uses of AI
DeepFakes, fakes can move faster than the truth can catch up
Undermining of democracy and privacy, oppressive surveillance
Generating fake comments
spam vs. anti-spam, fraud vs. anti fraud
AI and developing economies
AI will eliminate lower rung opportunities. The development of leapfrog opportunities will be required. Think how countries jumped to mobile phones, mobile payments, online education, etc.
US and china leading, but still a very immature space.
Use AI to strengthen country's vertical industries.
More public-private partnerships
invest in education
AI and Jobs
AI is automation on steroids.
Conditional basic income: provide a safety net but incentivize learning
Lifelong learning society
What is AI?
Building AI projects
Building AI in your company
AI and society
Case Study: Smart Speaker
- “Hey Device, tell me a joke”
- Steps (AI Pipeline):
- Trigger work/wakeword detection A) Hey device”? -> B) 0/1
- Speech recognition A) Audio -> B) “tell me a joke”
- Intent recognition A) Joke? vs, B) time?, music?, call?, weather?
- Execute joke
- Steps (AI Pipeline):
- Trigger work/wakeword detection A) Hey device”? -> B) 0/1
- Speech recognition A) Audio -> B) “Set timer for 10 minutes”
- Intent recognition A) “set timer for 10 minutes -> B) Timer
- Extract duration
- “Set timer for 10 minutes”
- “Let me know when 10 minutes is up”
- Start Timer with set duration
- Each function is a specialized piece of software.
- This requires companies to train users on what the speaker can, and can not do.
Case study: Self driving car
- Steps for deciding how to drive
- Car detection
- Pedestrian detection
- Car detection (supervised learning)
- Pedestrian detection (supervised learning)
- Motion Planning (SLAM – Simultaneous localization and mapping)
- Each function is a specialized piece of software.
Roles in AI teams
- Software Engineers (30% +)
- Machine Learning Engineer. focused on A -> B mapping
- Applied ML Scientist: Using State of the art to today’s problems
- Machine Learning Researcher. Extend the state-of-the-art in ML
- Data Scientist. Examine data and provide insights. Make presentation to team/executives. Some may be Machine Learning Engineer.
- Data Engineers. Organize data. Make sure data is saved in an easily accessible, secure, and in a cost effective way
- AI Product Manager. Help define what to build. What feasible and valuable
AI Transformation Playbook
- Execute pilot projects to gain momentum
- Success is more important than value
- Need to get the flywheel moving
- Show traction within 6 to 12 months (quiz said 6 to 10)
- Can be in-house or outsourced
- Can be under: CTO, CIO, CDO, or CAIO
- Have a central AI center of excellence. Matrix them in to start, untill understanding of AI is throughout the org
- CEO should provide funding to start, not from BU.
- You do this at step 4 to gain concrete experience, vs starting with an academic strategic approach to something so new.
- Leverage AI to create and advantage specific to your industry sector
- Design strategy aligned with the “Virtuous Cycle of AI”
- Better product -> More users -> More data -> [Repeat]
- Create a data strategy
- Strategic data acquisition
- Unified data warehouse/lake
- Create network effects and platform advantages
- In industries with “winner take all/most” dynamics, AI can be an accelerator
- Leverage classic frameworks as well. Low cost/ focus
- Consider humanity.
- AI can change a company and its products
- Investor relations. to properly value your company
- Government relations. to align on regulations.
- Consumer/use education
- Internal communications. to address questions and concerns.
AI pitfalls to avoid
- Expect it to do everything
- Hire 2-3 ML engineers and expect then to come up with use cases.
- expect it to work the first time\
- Don’t expect Traditional planning process to apply yo AI
- Don’t wait for a superstar, get going with what you have today.
- Be realistic
- Beginners should be linked with business
- Work with the AI team to develop new timelines, KPIs, etc
Taking your first steps
- Get friends to learn about AI
- Reading group
- Start brainstorming projects (no project too small)
- Hire a few ML/DS people to help
- Hire or appoint an AI leader
- Discuss with CEO/Board possibilities of AI transformation
- Will the company be more valuable, or more effective if we are good at AI.