As artificial intelligence automates more of Wall Street's analytical work, one of the industry's top executives is sounding an alarm: banks risk creating a generation of financiers who can't think for themselves. Chris Churchman, who leads Goldman Sachs' Marquee platform for institutional clients, warns that outsourcing reasoning to AI models could trigger "cognitive atrophy" in the next generation of bankers and traders.

Why Are Banks Struggling to Balance AI With Human Development?

The tension is real and immediate. Wall Street has been aggressively embedding AI into trading, risk analytics, and client services to boost efficiency and profitability. But this speed-first approach may come at a hidden cost: the loss of tacit knowledge that separates seasoned professionals from junior employees.

"There's a huge danger here that in the era of AI, we outsource our reasoning to these models, and we have cognitive atrophy that stops us being able to reason from first principles ourselves," said Chris Churchman, who leads Goldman's digital platform for institutional clients called Marquee.

Chris Churchman, Head of Marquee, Goldman Sachs

Churchman draws a parallel to how modern technology has eroded other human skills. Just as GPS navigation eliminated the need for map-reading and memorization, AI could strip away the analytical muscle that junior traders and bankers develop through hands-on experience. The problem is that finance isn't like navigation; it requires judgment calls in high-stakes, uncertain situations where reasoning from first principles can mean the difference between profit and catastrophic loss.

How Can Banks Preserve Talent Development While Adopting AI?

Churchman and other industry leaders are grappling with a practical framework for integrating AI without sacrificing the apprenticeship model that has defined Wall Street for decades. The challenge isn't technical; it's cultural and organizational.

• Preserve Human Decision-Making in High-Stakes Scenarios: Systems must be designed so that employees remain active decision-makers in complex, uncertain situations rather than becoming passive operators who simply execute AI recommendations without understanding the underlying logic.
• Maintain Hands-On Learning for Junior Staff: Junior traders traditionally learn by fielding client pricing requests under the supervision of experienced risk takers. While this work could be automated, doing so risks eliminating the training ground where the next generation of senior traders develops their intuition and judgment.
• Protect Tacit Knowledge Transfer: Much of what makes a great banker or trader cannot be written down or codified into an algorithm. Banks need to ensure that senior professionals actively mentor junior staff and that AI tools augment rather than replace this mentorship dynamic.

Churchman acknowledged that even Goldman Sachs, one of the world's most sophisticated investment banks, hasn't yet "figured out" how to manage this transition. The firm is still determining how to balance AI adoption with the need to preserve the institutional knowledge and reasoning capabilities that define its competitive advantage.

The stakes extend beyond individual firms. Wall Street has already begun examining ways to use AI to reduce the ratio of junior bankers to senior employees, potentially shrinking the pipeline of future talent. If this trend accelerates without deliberate safeguards, the industry could face a skills crisis within a decade.

What Technical Challenges Are Slowing AI Adoption in Finance?

Beyond the talent question, Goldman is confronting a fundamental technical hurdle: accuracy. In consumer-facing AI applications like chatbots, occasional errors are tolerable. In high finance, the tolerance for mistakes is near zero. A miscalculation in risk analytics or trade execution can cost millions of dollars in seconds.

"When we challenged it hard, at least it was honest. It was like, 'Look, in the end, I'm better at sounding thorough than being thorough,'" said Churchman, describing how Goldman's AI platform acknowledged its own limitations during testing.

Chris Churchman, Head of Marquee, Goldman Sachs

This candid admission from an AI system highlights a critical vulnerability: large language models can sound authoritative while being factually wrong. For Marquee, which is currently available only to Goldman employees, the firm is working to ensure that every AI-generated answer is 100% factual and can be audited in real time. This is a far higher bar than most consumer AI applications face.

The broader implication is clear: Wall Street's AI revolution will not be a simple plug-and-play adoption of off-the-shelf models. Banks will need to invest heavily in customization, validation, and governance frameworks to ensure that AI enhances rather than undermines their core business and talent development. The firms that get this balance right will likely emerge stronger; those that don't risk sacrificing long-term competitive advantage for short-term efficiency gains.