A Goldman Sachs partner leading one of the bank’s major artificial intelligence initiatives has warned that the rapid adoption of AI across Wall Street could create an unexpected long-term risk: weakening the reasoning skills of the next generation of bankers and traders.
Chris Churchman, who leads Goldman Sachs’ Marquee platform for institutional clients, said financial professionals could become too dependent on AI systems to analyze problems and develop conclusions.
“There’s a huge danger here that in the era of AI, we outsource our reasoning to these models,” Churchman said during Goldman Sachs’ “Exchanges” podcast, according to a transcript provided to CNBC.
His comments come as banks accelerate the use of AI across trading, research, investment banking and other financial operations. Goldman Sachs has already expanded access to generative AI tools across the firm as part of a broader effort to improve productivity.
The technology could make financial institutions more efficient. However, Churchman argues that removing too much routine work could also eliminate an important part of how young professionals develop judgment.
AI Could Change How Wall Street Trains Young Bankers
Churchman compared the potential impact of AI with previous technologies that reduced people’s need to rely on certain cognitive skills.
Navigation systems, for example, have reduced the need for people to memorize routes or develop strong spatial awareness. AI could create a similar effect in finance if employees increasingly delegate analytical work to algorithms.
“Reasoning is still important,” Churchman said. “You still need to reason about [problems] and structure it into an argument, and now we’re delegating reasoning.”
The concern is particularly relevant for junior bankers and traders.
Wall Street has traditionally relied on an apprenticeship model. Young employees perform routine tasks under the supervision of experienced professionals. Over time, those experiences help them develop judgment, intuition and an understanding of how markets behave.
AI could automate many of those early-career responsibilities.
That may improve productivity in the short term. But it could also create a talent problem several years later if fewer junior employees have experienced the situations that help them become senior decision-makers.
Churchman said banks need to preserve what he described as tacit knowledge.
“You learn by doing, and a lot of knowledge is tacit, it was never written down,” he said.
Goldman Sachs’ own technology platforms illustrate how deeply automation is already embedded in the firm’s financial operations. Its Marquee platform gives institutional clients access to market insights, analytics, execution tools and data services. Goldman Sachs Marquee platform
That technology can help professionals make faster decisions. But Churchman argues that speed should not come at the expense of developing the ability to reason independently.
Trading provides a clear example.
Junior traders often learn by handling client pricing requests while experienced traders supervise their decisions. That process teaches them how to assess risk and make judgments under pressure.
AI can automate much of the pricing process.
But Churchman questioned what happens if future senior traders never develop the underlying skills because an AI system handled those decisions for them.
“We can absolutely automate that,” he said, “but then do we get the senior traders that fully understand?”
The question extends beyond Goldman Sachs.
Banks across Wall Street are facing the same dilemma: how to capture the productivity benefits of AI without removing the human experiences that create expertise.
Goldman Sachs Faces a New AI Reliability Challenge
Churchman also discussed another major challenge: making AI systems reliable enough for high-stakes financial applications.
Marquee provides institutional clients with access to Goldman Sachs market data, research, risk analytics and execution services. The platform has evolved into a broader digital ecosystem that connects Goldman Sachs’ proprietary data and analytical capabilities with institutional workflows. Goldman Sachs Marquee MarketView
Goldman Sachs is also developing AI capabilities for its financial platforms. Churchman said the Marquee AI platform discussed in the podcast was available to Goldman employees at the time of the interview.
The technical challenge is not simply getting an AI model to produce an answer.
The bigger problem is determining whether that answer is accurate, explaining how it was generated and ensuring that the result can be audited.
That standard is particularly important in financial markets.
A consumer asking an AI chatbot a general question may tolerate an incorrect response. A trader, portfolio manager or risk officer cannot necessarily do the same.
A factual error can contribute to a financial loss, an incorrect trade or a flawed risk assessment.
The Federal Reserve has also highlighted the importance of responsible AI adoption in banking, noting that financial institutions must consider both the potential benefits and risks of generative AI. Federal Reserve analysis of AI in banking
Churchman said Goldman Sachs discovered an unusual limitation while testing its AI system.
When challenged extensively, the model effectively acknowledged that it could produce responses that sounded more rigorous than they actually were.
That observation highlights one of the central problems with generative AI in professional environments.
An AI system can produce a detailed explanation without necessarily reaching the correct conclusion.
For finance, that distinction matters.
An answer that sounds confident can be more dangerous than an obvious error because users may be less likely to question it.
Financial institutions therefore need systems that can provide useful answers while allowing important decisions to be independently reviewed.
Goldman Sachs has simultaneously continued to study how AI could reshape financial markets and investment processes. Its research has examined how generative AI tools can change systematic investing and the way professionals evaluate financial information. Goldman Sachs research on generative AI and investing
The Wall Street AI Revolution Has a Human Cost
The financial industry has strong incentives to automate.
Banks process enormous amounts of information and perform many repetitive tasks that are well suited to AI. Automating those processes can reduce costs, accelerate analysis and allow employees to focus on higher-value activities.
Goldman Sachs has already described AI as an important part of its effort to increase productivity and improve workflows. Its 2025 annual report also highlighted the firm’s broader investment in artificial intelligence and technology.
But Churchman’s warning points to a less obvious consequence.
If AI removes too many entry-level responsibilities, banks could save money today while weakening their talent pipeline for tomorrow.
That creates a difficult balancing act.
Banks could deliberately preserve certain tasks for junior employees even when AI can perform them more efficiently. The purpose would not necessarily be productivity. It would be training.
Employees could also be required to analyze selected problems independently before consulting AI systems. This could help them develop their own reasoning while still benefiting from automation.
Another approach would be to use AI as a decision-support system rather than a replacement for human judgment.
Under that model, AI could identify patterns, summarize information and suggest possible solutions. Human professionals would remain responsible for questioning the output and making the final decision.
That distinction becomes particularly important when uncertainty is high.
Churchman argued that employees should remain the decision-makers in situations where the consequences are significant and the available information is incomplete.
Goldman Sachs has not yet completely solved how that transition should work.
The financial industry is moving quickly toward AI because the economic incentives are clear. But some of the most valuable skills on Wall Street may not be the easiest ones to automate.
Reasoning, intuition, judgment and the ability to recognize when something does not make sense are developed through experience.
If AI performs too much of that work, banks could eventually face a paradox: increasingly sophisticated technology operating alongside a less experienced human workforce.
For Goldman Sachs and its competitors, the challenge will be finding the point where AI makes bankers more capable without allowing the technology to become a substitute for thinking.
