On paper, it sounds great. After all, who doesn’t want their pension managed by an egoless digital supermind? But just before we completely loosen the reins, a comprehensive study published recently presents us with a difficult sight. He asks a question that is a bit scary to ask: Is it possible that the learning process of these robots is actually being carried out… at our expense?
To understand the problem, you need to understand how the new machines learn. Unlike the old software that worked according to rigid rules (“If the stock goes down 2%, sell”), today’s algorithms are autonomous agents. They don’t get instructions, they get a goal: to make money. the way there? They have to find out for themselves.
How do they find out? through an investigative process. Imagine a child learning to walk and falling over and over again until he succeeds. It’s cute when the kid falls on the carpet in the living room. But in the capital market, when the algorithm decides to “run an experiment” to see how the market will react, it does so with real money.
In a game of chess, the computer can play a million games against itself without it costing anyone a dime. In the stock market, an algorithmic “experiment”—for example, a flash sale to see if there are buyers—causes real price swings. And are they? They are the “noise” that affects the value of all of our bags.
The research found an amazing thing: one robot is nice, but when there are lots of robots all trying to learn from each other at the same time, a contradiction and a collision arise. Algorithm A does an experiment, Algorithm B thinks it’s real information and reacts in panic, and suddenly we got a “flash avalanche” for no real economic reason. The price stops reflecting the value of the company and becomes an arena of struggle between machines.
We are at a critical point. Technology is here to stay, and it will indeed bring with it tremendous productivity in the future. But right now, we are in the dangerous transition phase. Without smart regulation to limit these experiments in the live trading environment, the capital market may lose its role as a barometer of the economy.
Don’t panic, but be vigilant. When you see sharp and inexplicable fluctuations in your portfolio in the coming years, it may not be because the economy has changed, but simply because some robot decided it was time to learn a new lesson and you are the ones paying the tuition.