Artificial Intelligence is the new PC

Artificial Intelligence is making breakthrough advancements in image recognition, natural language processing, robotics, and machine learning in general. The company DeepMind, a leader in AI research, has created AlphaZero in 2018, an AI program that has reached superhuman performance in many games including the game of Go, and more recently in 2020, AlphaFold that solved a protein folding problem that has preoccupied researchers for 50 years. For Stanford Professor Andrew Ng, AI is the new electricity.

A more useful comparison would be the advent of the personal computer, in particular, the IBM PC in 1981, and its first killer application, the spreadsheet software Lotus 123. IBM didn’t introduce the first computer. Home computers were already available for hobbyists since 1977 from companies such as Commodore, Tandy, and Apple. The Apple II with Visicalc was especially already very popular but the IBM PC was the first affordable personal computer enthusiastically adopted by the business community.  

Figure 1. IBM PC

Figure 2. Lotus 123

The novel spreadsheet software allowed flexible free-form calculations, the automation of calculations, the use of custom functions, graphics, references, and data management. Excel, the dominant spreadsheet software is still in use more than thirty years after its first introduction (with more functionalities). Before the spreadsheets, people used calculators and reported results on paper. More intensive calculations were done with mainframe computers in a language such as FORTRAN and results were printed on paper.

Today, AI is the new PC. Not adopting AI is like forgoing the PC in 1981. The impact is already very profound among the native digital companies and should be as significant for the rest of the companies.

Today, business leaders need to think about an AI strategy as they have to think about their information technology strategy. Like the PC and the spreadsheet, they should expect all their employees to become at some point users of AI at work. As the home computer, AI is already present at home with personal assistants such as Amazon Alexa, on phones with Apple Siri, and the internet with Google. All these AI applications are now possible thanks to increasing computer power, the development of the cloud, the availability of big data, and the new machine deep learning paradigm.

The AI Strategy Handbook was written to help you adopt AI in your business strategy so that it creates a long-term sustainable competitive advantage for your customers, your company, your employees, and your investors.

Building Machines that Learn and Think Like People

MIT Prof. Josh Tenenbaum gave a talk on Building Machines that Learn and Think Like People at the ICML 2018. His insight is that it is possible to teach a machine to learn like a child by using:

  • Game engine intuitive physics
  • Intuitive psychology
  • Probabilistic programs
  • Program induction
  • Program synthesis

This agenda is more ambitious that the current state of machine learning though it resembles more old-style machine learning an there is no guarantee that it will succeed. Still it is refreshing that we can learn from young humans to teach machines.

Imitation Learning

At the ICML 2018 conference there was a very interesting tutorial on Imitation Learning by Yisong Yue and Hoang Le from CalTech. It is quite similar to Reinforcement Learning but with an expert that the machine wants to imitate by inferring a policy that links states to actions. Imitation Learning can be applied to sequential decision making problem made by humans or other algorithms.

There are different categories of Imitation Learning:

  • Behavioral Cloning which is supervised learning on the state-action pairs of the expert
  • Direct Policy Learning (Interactive Imitation Learning) with interaction with an expert
  • Inverse Reinforcement Learning which is reinforcement learning applied to an inferred reward function from demonstrations

Direct Policy Learning can use Sequential Learning Reduction algorithms such as Data Aggregations (DAgger) and Policy Aggregations (SEARN & SMILe).

According to the presenters Imitation Learning seems to be easier to implement that Reinforcement Learning. A limitation is that the machine cannot do better than the expert. The talk is here:


Reproducibility, Reusability, and Robustness in Deep Reinforcement Learning

Mc Gill Professor Joelle Pineau has an insightful presentation on reproducibility in machine learning and especially in deep reinforcement learning. This is a general trend in science that some results sometimes cannot be fully reproduced. In deep reinforcement learning, there is a stochastic component to the results such as the present value of future rewards. She observes that results can vary for reasons that should not matter such as picking up a random seed (to generate random variables) and that the implementation of base cases by different researchers can yield different outcomes. Making the code and the data available for other researchers to reproduce paper results could alleviate some of these problems. She has introduced the Reproducibility Challenge that could be adopted by other scientific conferences.

Is Interpretability Necessary for Machine Learning?

At the NIPS 2017 conference there was a fascinating debate on the necessity of interpretability in machine learning. Without interpretability, mistakes can be made for instance when correlation is just used as a proxy for causation as Rich Caruana illustrates with a medical example. Yann Lecun on the other hand thinks that it is not necessary, it just needs to work. According to Lecun, people are not really interested in looking into the intimate details of a machine learning model, they just want their models to work. Kilian Weinberger argues that between an interpretable model with high error rate and a non-interpretable model with low error rate, people would choose the latter.

Interpretability is closer to what an economist would require, having a model that can be explained with model parameters that can be estimated and interpreted. If the model cannot be explained there is always a risk of capturing spurious correlations, having an omitted variable bias (when an important explanatory variable is missing) or endogeneity problems (when an explanatory variable is correlated with the error term) and really be wrong. At the same time, for real life applications such as forecasting or medical diagnostics, model accuracy is probably more important.

Without interpretability, there is a risk that a machine learning model will make mistakes that a human with “common sense” would not make. A more serious risk that the actual error rate will be higher on real world data when it is deployed in production (The model is wrong). Economists use interpretable economic models to limit this risk. In the current state of machine learning, there seems to be a trade-off between interpretability and accuracy (or effectiveness). Some promising approaches have been suggested to make machine learning models more interpretable for instance by approximating them by simpler local models such LIME. You can read this post.

More rigorous testing of the models can also be used and sometimes confronting the models with “common sense” or the current state of knowledge of the field can be useful.  In domains where machine ave reached superhuman skills (think of Alpha Go) the latter approach might not however be possible.

We encourage you to follow this debate:


What Can Machine Learning Do? Workforce Implications

This is an economist talk about the implication of machine learning on the workforce by Professor Erik Brynjolfsson at the ICLR 2018 conference. The effect of technology on jobs seems to be very strong on unskilled workers and has been reinforcing inequality. He provides some indications on jobs that will survive AI such as massage therapist and anything that requires human/social interactions.

Arxiv Sanity

A great website  to sort out research papers in Machine Learning is arxiv sanity. This was developed by Andrej Karpathy now at Tesla. Look at the introductory video:

You can save papers that you like and look for similar papers as ranked by their tf-idf statistics. What is missing is probably a social score of these papers such as which paper is popular at the moment though there are the top saved papers which can serve as a proxy for popularity.