Artificial Intelligence (AI) will transform our lives, and prediction machines will transform our understanding of AI. The book focuses on what managers need to know about the AI revolution, while taking a grounded and realistic perspective of such technology. The authors use principles of economics and strategy to understand how firms, industries and management will be transformed by AI.
These are my main takeaways out of this book, written by Ajay Agrawal, Joshua Gans and Avi Goldfarb:
- Artificial Intelligence (AI) is everywhere. It’s in our phones, cars, shopping experiences, romantic matchmaking, hospitals, banks, and all over the media.
- AI is a prediction technology – predictions are inputs to decision making, and economics provides a perfect framework for understanding the trade-offs underlying any decision. The new wave of AI does not actually bring us intelligence but instead a critical component of intelligence – prediction. Then, a key question regarding AI is ‘what predictions are important for your business?‘ – Prediction is a central input into decision-making.
- Prediction is the process of filling missing information. Prediction takes information you have, often called ‘data’, and uses it to generate information you don’t have.
- Before machine learning, multivariate regression provided an efficient way to condition on multiple things, without the need to calculate dozens, hundreds or thousands of conditional averages. However, being precisely ‘perfect on average’ can mean being actually wrong each time.
- Prediction machines rely on data – more and better data leads to better predictions. But data can be costly to acquire. Thus, the investment involves a trade-off between the benefit of more data and the cost of acquiring it.
- Prediction is not a decision. Making a decision requires applying judgment to a prediction and then acting. In consequence, advances in machine prediction mean that we have to examine the anatomy of a decision. By breaking out a decision into elements, we can think clear about which parts of human activities will diminish in value and which will increase as a result of enhanced machine prediction
- Anatomy of a task: when someone (or something) makes a decision, they take input data from the world that enables a prediction. That prediction is possible because of training about relationships between different types of data to identify which is most closely associated with a situation. While combining the prediction with judgment on what matters, the decision maker can then choose an action. The action leads to an outcome (which has an associated reward or payoff). The outcome is a consequence of the decision and may also provide feedback to help improve the next prediction.
- The ‘AI canvas‘ is a useful tool to decompose tasks in order to understand the potential role of a prediction machine. This means identifying: action; prediction; judgment; outcome; input; training; and feedback. Fill out the canvas for every decision or task, and this will introduce discipline and structure into the process.
- Prediction machines don’t provide judgment. Only humans do, because only humans can express the relative rewards from taking different actions. Such payoffs are rarely obvious, and the process of understanding those payoffs can be time consuming and costly.
- Many decisions occur under conditions of uncertainty, and we need to determine the payoff for acting on wrong decisions, not just right ones. So, uncertainty increases the cost of judging the payoffs for a given decision.
- There are limits to the ability of machines to predict human judgment. The limits rely on lack of data. There is some data that humans have and machines do not, such as individual preferences. Such data has value, and companies currently pay to access it through discounts on using loyalty cards and free online services like Google and Facebook.
- When you employ a prediction machine, the prediction made must be communicated to the decision maker. But if the prediction leads directly to an obvious course of action (“no need to think“), then the case for leaving human judgment in the loop is diminished. If a machine can be coded for judgment and handle the consequent action relatively easy, then it makes sense to leave the entire task in the machine’s hands.
- AI tools can change work flows in two ways. First, they can render tasks obsolete and therefore remove them from work flows. Second, they can add new tasks. This may be different for every business and every work flow.
- The implementation of AI tools generates four implications for jobs: augment jobs; contract jobs; the reconstitution of jobs; or a shift on the emphasis of specific skills required for a particular job.
- Selling predictions – Google, Facebook, Microsoft, and handful of other companies have particularly useful data on consumer preferences online. Rather than only sell data, they go a step further to make predictions for advertisers.
- Unique data is important for creating strategic advantage. If data is not unique, it is hard to build a business around prediction machines. Without data, there is no real pathway to learning, so AI is not core to your strategy. AI will increase incentives to own data.
- A key strategic choice is determining where your business ends and another begins – deciding on the boundary of the firm. Uncertainty influences this choice. Because prediction machines reduce uncertainty, they can influence the boundary between your organisation and others.
- Experience is the new scarce resource – Navigation app Waze collects data from others Waze users to predict the location of traffic problems. It can find the fastest route for you personally. If that were all it was doing, there would be no issue. However, prediction alters human behaviour, which is what Waze is designed to do. When the machine receives information from a crowd, its predictions may be distorted by that fact.
- Cheap means everywhere: when the price of something fails, we use more of it – and that is precisely happening now with AI. Amazon illustrates such situation: its current business model is shopping-then-shipping. However, when the accuracy of its recommendation engine (currently at 5%) will improve, it will be able to move to a shipping-then-shopping business model, while sending items Amazon will know you want to buy.
- Will inequality get worse because of AI? – The rise of AI presents three particular trade-offs to society: a) productivity vs distribution – the problem isn’t wealth creation, it is distribution; b) innovation vs competition – businesses have greater incentives to build prediction machines if they have more control, but along with its scale economies, this may lead to monopolisation; and c) performance vs privacy – since AI performs better with more data.
Prediction is at the heart of making decisions under uncertainty, and our businesses and personal lives are riddled with such decisions. Given that uncertainty constraints strategy, better prediction creates opportunities for new business structures and strategies to compete.
The impact of AI will be profound, but the economic framework for understanding it is surprisingly simple.