There has been no lack of interest in the capabilities of AI, along with related concerns about bias, data validity, costs, and employee resistance. But we may be missing the most important point when it comes to the ultimate impact of AI, argues one prominent proponent of AI. That is, we are starting to outsource a large part of human decision-making to machines, which could have unexpected implications – other than just making cheaper predictions.
It’s time to start looking at AI not from a technical perspective, but from an economic one, says Ajay Agrawal, a professor at the University of Toronto and co-author of Strength and Prediction: The Disruptive Economics of Artificial Intelligence. Agrawal recently shared his views on the next wave of AI at A accident Hosted at the University of British Columbia’s Green College. AI is moving to its next stage – moving up the food chain for decision-making. This is where AI is moving from the sidelines to a more central role in the economy, he says.
In general, there has been disappointment with AI, as it doesn’t seem to deliver the miracles promised at first, he adds, noting that many “things seem to be much less affected than we thought”. Even productivity growth continues to decline. At the same time, Agrawal continues, AI is still a work in progress, and we’re just beginning to see it unfold.
Agrawal asserts that it is time to take an “economist’s point of view” in AI. A computer scientist or engineer will talk about artificial intelligence in terms of developments in neural networks. But if you ask an economist what’s going on with AI, they’ll describe it as a reduction in the cost of forecasting. As AI gets better and better, it effectively makes prediction cheaper and cheaper.” This is important because “we use forecasting everywhere. Forecasting is embedded in all kinds of things that you might not think about predicting — for example, self-driving.”
Agrawal says that decision-making, which is the source of financial and political power in an economy, consists of two components: forecasting and judgment. These two functions are separated in AI systems – humans retain judgment, but shift prediction to AI. “We are constantly doing some form of probability evaluation and judgment evaluation whether we realize it or not,” he says. “The rise of AI is shifting one of those components — prediction — from humans to machines. We are outsourcing the prediction part to the machine.”
So far, AI has focused on point solutions – copying text, detecting errors in production lines, etc. “We picked all the dangling fruits for all the point solutions where you only get a prediction, prediction does a simple procedure,” Agrawal says. “Like a tool — attached to a cam — that predicts whether a tooth on a digger in the mining process should break. This is a point solution, an expectation that leads to a specific action. It does not affect anything else in the process.”
AI begins to see greater value “when you start to build a completely independent system, where one expectation of one decision affects many other decisions,” Agrawal points out. “From an economic perspective, we’re in the realm of game theory, where if we change one decision, how does it affect all other decisions?”
Transferring the predictive aspect of decisions to machines can be an amazing experience when brought up. Agrawal says AI opens the door to a “new decision boom”. “Many of these decisions are new because we previously hid them through rules, insurance, and over-engineering,” he says. “We have done such a good job of hiding them that we have long forgotten that they were there at all. AI reveals these long-hidden decisions.”
This is more than just an exercise in creativity – it means strength. “Decision making confers power; changes in decision making can lead to changes in power,” he says. “Centralization or decentralization of decision-making will enhance or distribute power.”
This means a shift across the economy, Agrawal says. AI is arguably the first tool in human history that learns while using it. And the more you use it, the smarter it gets because every time you use it.”