Prediction Machines: The Simple Economics of Artificial Intelligence

by Ajay Agrawal

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.

If economists are good at one thing, it is cutting through hype. Where others see transformational new innovation, we see a simple fall in price. But it is more than that. To understand how AI will affect your organization, you need to know precisely what price has changed and how that price change will cascade throughout the broader economy.

The rise of the internet was a drop in the cost of distribution, communication, and search. Reframing a technological advance as a shift from expensive to cheap or from scarce to abundant is invaluable for thinking about how it will affect your business.

insofar as a computer was concerned, she wrote, it “had no pretensions to originate anything. It can do whatever we know how to order it to perform. It can follow analysis; but it has no power of anticipating any analytical relations or truths.”7 Despite all the hype and the baggage that comes with the notion of AI, what Alan Turing later called “Lady Lovelace’s Objection” still stands. Computers still cannot think, so thought isn’t about to become cheap.

What will new AI technologies make so cheap? Prediction. Therefore, as economics tells us, not only are we going to start using a lot more prediction, but we are going to see it emerge in surprising new places.

The same goes for prediction. Prediction is being used for traditional tasks, like inventory management and demand forecasting. More significantly, because it is becoming cheaper it is being used for problems that were not traditionally prediction problems.

Kathryn Howe, of Integrate.ai, calls the ability to see a problem and reframe it as a prediction problem “AI Insight,” and, today, engineers all over the world are acquiring it.

When prediction is cheap, there will be more prediction and more complements to prediction. These two simple economic forces drive the new opportunities that prediction machines create.

Some AIs will affect the economics of a business so dramatically that they will no longer be used to simply enhance productivity in executing against the strategy; they will change the strategy itself.

What does this mean for strategy? First, you must invest in gathering intelligence on how fast and how far the dial on the prediction machines will turn for your sector and applications. Second, you must invest in developing a thesis about the strategic options created from turning the dial.

Prediction facilitates decisions by reducing uncertainty, while judgment assigns value. In economists’ parlance, judgment is the skill used to determine a payoff, utility, reward, or profit. The most significant implication of prediction machines is that they increase the value of judgment.

Innovations in prediction technology are having an impact on areas traditionally associated with forecasting, such as fraud detection.

Prediction is “intelligence” in the espionage sense of “obtaining of useful information.”10 Machine prediction is artificially generated useful information. Intelligence matters. Better predictions lead to better outcomes,

What does regression do? It finds a prediction based on the average of what has occurred in the past.

regression models aspire to generate unbiased results, so with enough predictions, those predictions will be exactly correct on average.

Being precisely perfect on average can mean being actually wrong each time. Regression can keep missing several feet to the left or several feet to the right. Even if it averages out to the correct answer, regression can mean never actually hitting the target.

Unlike regression, machine learning predictions might be wrong on average, but when the predictions miss, they often don’t miss by much. Statisticians describe this as allowing some bias in exchange for reducing variance.

Machine learning models are particularly good at determining which of many possible variables will work best and recognizing that some things don’t matter and others, perhaps surprisingly, do. Now, an analyst’s intuition and hypotheses are less important. In this way, machine learning enables predictions based on unanticipated correlations, including that housing prices in Las Vegas, Phoenix, and Miami might move together.

Moreover, the process of machine learning involves searching for a solution that tends to minimize errors.

In his book On Intelligence, Jeff Hawkins was among the first to argue that prediction is the basis for human intelligence. The essence of his theory is that human intelligence, which is at the core of creativity and productivity gains, is due to the way our brains use memories to make predictions: “We are making continuous low-level predictions in parallel across all our senses. But that’s not all. I am arguing a much stronger proposition. Prediction is not just one of the things your brain does. It is the primary function of the neocortex, and the foundation of intelligence. The cortex is an organ of prediction.”

Data scientists have excellent tools for assessing the amount of data required given the expected reliability of the prediction and the need for accuracy. These tools are called “power calculations” and tell you how many units you need to analyze to generate a useful prediction.

data might be most valuable if you have more and better data than your competitor. Some have argued that more data about unique factors brings disproportionate rewards in the market.

Humans and machines both have failings. Without knowing what they are, we cannot assess how machines and humans should work together to generate predictions.

Prediction proves so difficult for humans in this situation because of the complexity of the factors that might explain crime rates. Prediction machines are much better than humans at factoring in complex interactions among different indicators. So, while you might believe that a past criminal record may mean that a defendant is a bigger flight risk, the machine may have discovered that is only the case if the defendant has been unemployed for a certain period of time. In other words, the interaction effect may be the most important, and as the number of dimensions for such interactions grows, humans’ ability to form accurate predictions diminishes.

In contrast to machines, humans are sometimes extremely good at prediction with little data.

Faced with unknown unknowns, both humans and machines fail.

Determining whether an action causes an outcome requires two predictions: first, what outcome will happen after the action is taken, and second, what outcome would have happened had a different action been taken. But that’s impossible. You will never have data on the action not taken.16

reverse causality remains a challenge for prediction machines. This issue appears frequently in business, too. In many industries, low prices are associated with low sales. For example, in the hotel industry, prices are low outside the tourist season, and prices are high when demand is highest and hotels are full. Given that data, a naive prediction might suggest that increasing the price would lead to more rooms sold.

Sometimes, the combination of humans and machines generates the best predictions, each complementing the other’s weaknesses.

How does such collaboration translate into a business environment? Machine prediction can enhance the productivity of human prediction via two broad pathways. The first is by providing an initial prediction that humans can use to combine with their own assessments. The second is to provide a second opinion after the fact, or a path for monitoring.

For a human–prediction machine pair to generate a better prediction requires an understanding of the limits of the human and the machine.

it is important to understand the weaknesses of both humans and machines and combine them in a way that overcomes these flaws.

We describe a taxonomy of prediction settings (i.e., known knowns, known unknowns, unknown knowns, and unknown unknowns) that is useful for anticipating the appropriate division of labor.

Prediction machines scale. The unit cost per prediction falls as the frequency increases. Human prediction does not scale the same way. However, humans have cognitive models of how the world works and thus can make predictions based on small amounts of data. Thus, we anticipate a rise in human prediction by exception

But a prediction is not a decision. Making a decision requires applying judgment to a prediction and then acting. Before recent advances in machine intelligence, this distinction was only of academic interest because humans always performed prediction and judgment together. Now, advances in machine prediction mean that we have to examine the anatomy of a decision.

Prediction machines will have their most immediate impact at the decision level. But decisions have six other key elements (see figure 7-1). When someone (or something) makes a decision, they take input data from the world that enables a prediction. That prediction is possible because training occurred about relationships between different types of data and which data is most closely associated with a situation. 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. It is needed to provide a complete picture. The outcome may also provide feedback to help improve the next prediction.

Judgment involves determining what we call the “reward function,” the relative rewards and penalties associated with taking particular actions that produce particular outcomes.

a prediction is not a decision—it is only a component of a decision. The other components are judgment, action, outcome, and three types of data (input, training, and feedback).

The value of substitutes to prediction machines, namely human prediction, will decline. However, the value of complements, such as the human skills associated with data collection, judgment, and actions, will become more valuable.

Judgment involves determining the relative payoff associated with each possible outcome of a decision, including those associated with “correct” decisions as well as those associated with mistakes. Judgment requires specifying the objective you’re actually pursuing and is a necessary step in decision making. As prediction machines make predictions increasingly better, faster, and cheaper, the value of human judgment will increase because we’ll need more of it. We may be more willing to exert effort and apply judgment to decisions where we previously had chosen not to decide (by accepting the default).

As AI takes over prediction, humans will do less of the combined prediction-judgment routine of decision making and focus more on the judgment role alone.

With better prediction come more opportunities to consider the rewards of various actions—in other words, more opportunities for judgment. And that means that better, faster, and cheaper prediction will give us more decisions to make.

The broader point for decisions is that objectives rarely have only a single dimension. Humans have, explicitly and implicitly, their own knowledge of why they are doing something, which gives them weights that are both idiosyncratic and subjective.

As prediction machines provide better and cheaper predictions, we need to work out how to best use those predictions. Whether or not we can specify judgment in advance, someone needs to determine the judgment. Enter reward function engineering, the job of determining the rewards to various actions, given the predictions that the AI makes. Doing this job well requires an understanding of the organization’s needs and the machine’s capabilities.

Machines are bad at prediction for rare events. Managers make decisions on mergers, innovation, and partnerships without data on similar past events for their firms. Humans use analogies and models to make decisions in such unusual situations. Machines cannot predict judgment when a situation has not occurred many times in the past.

in their book Reengineering the Corporation, argued that to use the new general-purpose technology—computers—businesses needed to step back from their processes and outline the objective they wanted to achieve. Businesses then needed to study their work flow and identify the tasks required to achieve their objective and only then consider whether computers had a role in those tasks.

In deciding how to implement AI, companies will break their work flows down into tasks, estimate the ROI for building or buying an AI to perform each task, rank-order the AIs in terms of ROI, and then start from the top of the list and begin working downward.

Every task has a group of decisions at its heart, and those decisions have some predictive element.

In advising them, we found it useful to separate the parts of a decision into each of its elements (refer to figure 7-1): prediction, input, judgment, training, action, outcome, and feedback.

Companies often find themselves having to go back to basics to realign on their objectives and sharpen their mission statement as a first step in their work on their AI strategy.