When AI's set Prices

Nowadays, AI is used by companies to decide the price of goods and how/when to vary them. With no humans in the equation, this should avoid “price collusion”, right? (That’s legal speak for competing companies working together to keep prices higher than they would have been otherwise).

Economists played off some of the AI algorithms against each other in a simulated environment. Guess what? The shape of the graph of prices over time matches what you’d see if price collusion is being practiced! Worse, the study found that this pattern occurs even with different AI’s in the mix:

“Even when the active firms are three or four in number, when they are asymmetric, and when they operate in a stochastic environment.”

And here’s the kicker. Guess what shows up in the logs the AI’s create of their activities, which could be used later by regulators to check for collusion?

“What is most worrying is that the algorithms leave no trace of concerted action – they learn to collude purely by trial and error, with no prior knowledge of the environment in which they operate, without communicating with one another, and without being specifically designed or instructed to collude.”

R. Preston McAfee, an academic economist who has worked at Yahoo, Google and Microsoft as an economist, says AI’s are making life very hard for regulators:

“As companies come more and more to be run by what amount to black box mechanisms, the government needs more capability to deconstruct what those black box mechanisms are doing.”

As if that doesn’t sound hard enough, things are even worse, says McAfee:

“This, by the way, is a nontrivial thing. It's not like a flight recorder; it's distributed among potentially thousands of machines, it could be hundreds of interacting algorithms, and there might be hidden places where thumbs can be put on the scale.”

And as with all matters to do with money in the US, it will soon turn into a philosophical debate, as I can see from some comments on this topic:

“If two or more AIs are in cahoots,even tacitly, it is not different from the CEO of General Motors and the CEO of Ford being in cahoots.”

v/s

“But surely to be "in cahoots" they would have to actually communicate and coordinate their activity in some way?”

v/s

“So, two algorithms pick up the same signals and arrive at similar answers. Doesn't this just reflect well established economic principles?”

Or as Tyler Cowen says:

“No need to torture the data when you can torture the AIs. It’s going to be a strange world.”