Show summary Hide summary
A Goldman Sachs partner who leads one of the bank’s flagship artificial-intelligence initiatives warned that widespread adoption of AI on Wall Street could blunt the analytical skills of future bankers, according to a transcript of the firm’s “Exchanges” podcast provided exclusively to CNBC. His remarks underscore a growing tension between automation gains and the traditions that train junior staff to make judgment calls.
A threat to on-the-job learning
Chris Churchman, who heads Goldman’s digital platform for institutional clients, Marquee, said there is a real risk that firms will “outsource our reasoning” to machine models and suffer what he called cognitive atrophy. He compared the effect to how modern conveniences have eroded older generations’ navigation and memorization skills.
Ultra-thin women smoking cigarettes are making a comeback as a fashion trend
Windows Memory Diagnostic explained: what the test does and how to read results
Bankers traditionally learn by doing: junior traders handle client pricing requests under the supervision of experienced colleagues, absorbing tacit skills that are seldom written down. Churchman argued those interactions teach intuitive judgment that algorithms cannot easily replicate.
That matters because, while automation can raise short-term productivity and profits, it may also undermine the apprenticeship culture that cultivates senior talent. CNBC reported last year that some Wall Street firms are exploring ways to use AI to reduce the ratio of junior bankers to senior staff—an outcome Churchman warned could weaken the industry’s future bench.
Preserving human judgment in high-stakes decisions
Churchman urged banks to design systems so employees remain the decision‑makers in situations of high uncertainty. “Reasoning is still important,” he said. “You still need to reason about [problems] and structure it into an argument, and now we’re delegating reasoning.”
The challenge, he added, is to retain the informal knowledge passed down through mentorship. “You learn by doing, and a lot of knowledge is tacit, it was never written down,” Churchman said, stressing the need to ensure the next generation acquires the same instincts as today’s top traders.
Churchman’s background includes running currency trading at UBS before joining Goldman in 2021. He is also co-chair of the firm’s Global Banking and Markets AI working group, a role that places him at the center of Goldman’s effort to balance automation with human expertise.
Accuracy and auditability remain technical hurdles
On the technical side, Churchman described a core difficulty in deploying AI for clients: ensuring outputs are factual and auditable. Marquee, which provides market data, research, risk analytics and trade execution services to hedge funds and other institutional users, is already being adapted with AI features—but the AI-enabled version is currently available only to Goldman employees.
In finance, the margin for error is minimal. Unlike consumer chatbots that include broad disclaimers, institutional systems must be provably accurate and traceable. Churchman said the firm’s development work revealed that the software could acknowledge its limitations. “When we challenged it hard, at least it was honest,” he said. “It was like, ‘Look, in the end, I’m better at sounding thorough than being thorough.'”
That candor points to a broader trade-off: AI can automate routine tasks and scale services, but firms must decide how to preserve the human experience that builds judgment and trust in markets.
