JPMorgan Chase is keeping score on which of its engineers use AI.
The bank built internal dashboards that track individual usage of AI coding tools across the company, and they're already live. One dashboard shows nearly 70,000 employees given access to GitHub Copilot, the bank's AI coding assistant. About 24,000 show up as active. The rest aren't using it.
JPMorgan framed the tracking as a way to figure out where people need more training and support. But Global CIO Lori Beer put it more directly. She said the productivity gains from AI were "significant enough to change the internal conversation from 'should engineers use AI' to 'why aren't all engineers using AI.'"
The numbers behind that shift are hard to dismiss. JPMorgan's AI tools have boosted engineer productivity by 10% to 20%. CEO Jamie Dimon told Bloomberg this week that the bank will be "hiring more AI people and probably less bankers in certain categories," calling current AI use across marketing, risk, fraud and document management "the tip of the iceberg." The company's tech budget is approaching $20 billion this year, up roughly $2 billion from 2025.
The rest of Wall Street is moving in the same direction. Citigroup made AI training mandatory for about 240,000 employees and says its tools save around 100,000 developer hours per week. Bank of America reports that AI now handles 30 billion client interactions. Morgan Stanley estimates that banks could see about 18% higher pretax income once AI is fully embedded in workflows. Every major bank is pushing adoption hard.
They just disagree on how.
Goldman Sachs CIO Marco Argenti has publicly rejected tracking AI use at the individual level, comparing it to watching only one player on a football field. Goldman measures team-level results instead, things like how fast ideas go from concept to production and how quickly engineering backlogs shrink. Using individual usage data as a scoreboard would "miss the forest for the trees," Argenti said. He reports that employee sentiment toward AI at Goldman has actually shifted from skepticism to genuine buy-in, and he credits that to people discovering the tools on their own rather than feeling watched while they learn.
A Cornell University study found the same pattern. Workers monitored by AI reported lower autonomy, more complaints, and higher intentions to quit compared with those monitored by humans. The negative effects disappeared when the tools were framed as developmental aids rather than automated judges.
Sameer Gupta, who leads AI in financial services at EY, told Business Insider he hadn't heard of banks tracking individual AI use before JPMorgan. "In general, people don't like to be tracked," he said. "Set aside AI for a second. If somebody is tracking how many total hours in a day you are on Teams and videoconferencing, it makes you uncomfortable."
And all of this is happening as this summer's new hires arrive on Wall Street already fluent in the AI tools that their future colleagues are being graded on learning.
JPMorgan and Goldman are running what amounts to a live experiment on the biggest open question in AI workforce strategy: does pressure or autonomy produce better adoption? JPMorgan built the leaderboard. Goldman built the culture. Both are spending billions and both need their people to actually use the tools. Within a couple years the productivity data will make one of them look right and the other look very wrong. If your company recently handed you an AI tool and started asking how often you open it, this experiment isn't hypothetical for you. It's already your Monday morning.
