Source: Cheng, Lin & Zhao (2025) — “Does Generative AI Facilitate Investor Trading? Early Evidence from ChatGPT Outages”
At first this sounds like a technical finance question. Then you look closer and realise it is also a question about attention, incentives, and whether people are using information or just being impressed by it.
One way to test whether investors are actually using a tool is to see what happens when the tool breaks. This paper exploits ChatGPT outage periods — when the platform experienced unscheduled downtime — as a natural experiment in investor AI dependence. The finding: during ChatGPT outages, stock trading volume declined significantly. The effect is strongest for firms that had corporate news released immediately before or during the outage — exactly the situations where investors would most benefit from AI assistance in processing new information. The effect is also stronger for firms held by transient institutional investors, who are the most active, information-sensitive traders. Going further, the researchers find that short-run price impact and return variance also declined during outages, consistent with reduced informed trading. When investors couldn’t use AI to process information efficiently, they traded less and moved prices less. Long-run price informativeness, meanwhile, shows a positive relationship with AI-assisted trading — suggesting that AI helps prices reflect information more accurately over time. What I find particularly striking is the specificity of when the effect appears. It’s not a generalised reduction in trading activity during outages. It’s concentrated among firms with recent news, which is the precise situation where AI would add most value.
So I would not read this as a neat technology story. It is a messy information story. The tools improve, but people still have to decide what deserves attention. Annoying, but true.