Prediction Machines: The Simple Economics of Artificial Intelligence

by Ajay Agrawal

The automation of tasks forces us to think more carefully about what really constitutes a job, what people are really doing.

We have highlighted three types of data—training, input, and feedback data. Training data is used to build a prediction machine. Input data is used to power it to produce predictions. Feedback data is used to improve it. Only the two latter types are needed for future use. Training data is used at the beginning to train an algorithm, but once the prediction machine is running, it is not useful anymore.

the race for value capture does not respect traditional business boundaries. Instead, it challenges the ownership of actions that might otherwise have been an advantage.

powerful AI tools may result in significant redesign of work flows and the boundary of the firm.

Prediction machines will increase the value of complements, including judgment, actions, and data.

The three ingredients we highlighted in the previous chapter suggest that AI might lead to strategic change. First, lower cost versus more control is a core trade-off. Second, that trade-off is mediated by uncertainty; specifically, the returns to control increase with the level of uncertainty. Major airlines balance lower cost and more control by optimizing the boundaries of where their own activities end and those of their partners begin. If a prediction machine could cut through this uncertainty, then the third ingredient would be present and the balance would shift. Airlines would contract more to their partners.

AI might enable machines to operate in more complex environments. It expands the number of reliable “ifs,” thus lessening a business’s need to own its own capital equipment,

Counterintuitively, better prediction increases the uncertainty you have over the quality of human work performed: you need to keep your reward function engineers and other judgment-focused workers in house.

For AI startups, owning the data that allows them to learn is particularly crucial. Otherwise, they will be unable to improve their product over time.

Google, through search, YouTube, and its advertising network, has rich data on user needs. It does not sell the data. However, it does, in effect, sell the predictions that the data generates to advertisers as part of a bundled service. If you advertise through Google’s network, your ad is shown to the users that the network predicts are most likely to be influenced by the ad. Advertising through Facebook or Microsoft yields similar results. Without direct access to the data, the advertiser buys the prediction.

Unique data is important for creating strategic advantage. If data is not unique, it is hard to build a business around prediction machines.

If the data resides with an exclusive or monopoly provider, then you may find yourself at risk of having that provider appropriate the entire value of your AI. If the data resides with competitors, there may be no strategy that would make it worthwhile to procure it from them. If the data resides with consumers, it can be exchanged in return for a better product or higher-quality service.

However, in some situations, you and others might have data that can be of mutual value; hence, a data swap may be possible. In other situations, the data may reside with multiple providers, in which case, you might need some more complicated arrangement of purchasing a combination of data and prediction.

Adopting an AI-first strategy is a commitment to prioritize prediction quality and to support the machine learning process, even at the cost of short-term factors such as consumer satisfaction and operational performance. Gathering data might mean deploying AIs whose prediction quality is not yet at optimal levels.

when you want to manage AI for a purpose core to your own business, no off-the-shelf solution is likely. You won’t need a user manual so much as a training manual. This training requires some way for the AI to gather data and improve.

There are no easy ways to overcome the trade-off that arises when prediction alters crowd behavior, thereby denying AI of the very information it needs to form the correct prediction.

experience is a scarce resource, some of which you need to allocate to humans to avoid deskilling.

While software has always been subject to security risks, with AI those risks emerge through the possibility of data manipulation. Three classes of data have an impact on prediction machines: input, training, and feedback. All three have potential security risks.

A seemingly easy solution to the problem of system-wide failure is to encourage diversity in the prediction machines you deploy. This will reduce the security risks, but at the cost of reduced performance. It might also increase the risk of incidental smaller failures due to a lack of standardization. Just as in biodiversity, the diversity of prediction machines involves a trade-off between individual and system-level outcomes.

Be aware of how your predictions differ across groups of people. Question whether your predictions reflect underlying causal relationships and if they are really as good as they seem to be. Balance the trade-off between system-wide risks and the benefit of doing everything a little bit better. And watch for bad actors who may query your prediction machines to copy them or even destroy them.