Goldman Sachs Warns AI Could Cause "Cognitive Atrophy"

A senior Goldman Sachs technology leader leading one of the bank's flagship AI projects has issued a blunt warning about the very technology he's building: outsourcing too much reasoning to AI risks permanently weakening the analytical skills of the next generation of financiers, according to CNBC's reporting on comments made during the firm's "Exchanges" podcast.

Chris Churchman, who leads Marquee, Goldman's digital platform giving hedge funds and other large clients access to the bank's market data, research, analytics, and execution tools, was direct about the risk. "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," Churchman said, according to Yahoo Finance's reporting on the podcast episode.

Churchman's concern centers on a specific, structural feature of how banking talent has traditionally developed. Junior traders learn by fielding client pricing requests under the direct supervision of experienced risk takers, absorbing tacit, unwritten knowledge through repetition and mentorship over years. "You learn by doing, and a lot of knowledge is tacit, it was never written down," Churchman said, according to CNBC's reporting. He posed the tension directly: "We can absolutely automate that, but then do we get the senior traders that fully understand?"

Churchman, who ran currency trading at UBS before joining Goldman in 2021, framed the challenge as one Goldman needs to actively design around, not simply accept as an unavoidable tradeoff. The bank needs "to make sure we don't lose that tacit and intuitive knowledge that some of our best people have today [and] to ensure the next generation have it too," he said.

A genuinely notable detail from Churchman's account involves the AI system itself. While developing Marquee's AI capabilities, currently available only to Goldman employees, the platform surprised the development team by volunteering its own limitations without being asked. "It was like, 'Look, in the end, I'm better at sounding thorough than being thorough,'" Churchman recounted, according to Yahoo Finance's reporting. He identified accuracy and traceability as the steepest technical obstacles the team faced, since finance carries a far narrower margin for error than the kind of casual mistakes consumer chatbot users routinely tolerate.

The Tension Goldman Is Trying to Navigate

Churchman's comments echo similar concerns raised elsewhere at Goldman. Bracha Cohen, a Goldman partner in technology within asset and wealth management engineering, has separately advised young computer science graduates to "build judgment and not just skills," arguing that "AI may automate execution, but it can't fully replace decision-making, systems thinking, and ethical reasoning," according to Business Insider's reporting on her comments. This connects to a broader pattern of AI-driven workforce anxiety we've tracked closely, including our coverage of growing worker backlash as companies increasingly cite AI in layoffs and Goldman's own separate research finding that AI's labor market impact remains concentrated specifically among entry-level workers and certain exposed industries.

Churchman's warning is worth taking seriously for any business currently automating entry-level or junior-level analytical work without a deliberate plan for how the next generation of senior talent will develop the underlying reasoning skills those roles traditionally taught. This tension isn't unique to banking. Any industry where junior employees historically learned complex judgment through repetitive, supervised practice faces a similar structural risk once AI absorbs that repetitive work entirely.

For business leaders designing AI adoption strategy, Churchman's framing offers a useful distinction worth applying broadly: the goal isn't simply automating tasks efficiently, it's ensuring employees remain the decision-makers in high-stakes, high-uncertainty situations rather than becoming passive observers of an AI system's output.

What did the Goldman Sachs partner say about AI and cognitive atrophy?
Chris Churchman, who leads Goldman's Marquee platform, warned that outsourcing too much reasoning to AI models risks causing "cognitive atrophy," weakening employees' ability to reason from first principles, particularly among junior bankers.

What is Goldman's Marquee platform?
Marquee is Goldman Sachs' digital platform giving hedge funds and other large institutional clients access to the bank's market data, research, analytics, and execution tools, with new AI features currently available only to internal Goldman employees.

Why is Wall Street's apprenticeship model specifically vulnerable to AI?
Junior traders and bankers have traditionally learned tacit, unwritten knowledge by handling supervised tasks like client pricing requests. If AI fully automates that repetitive work, the traditional path for developing senior-level judgment and expertise could be disrupted.

Goldman Sachs partner Chris Churchman warned that AI risks causing "cognitive atrophy" among junior bankers if employees outsource too much analytical reasoning to AI models, threatening Wall Street's traditional apprenticeship-based skill development. Churchman, who leads Goldman's Marquee platform, said the bank's own AI system surprised developers by unprompted admitting it is "better at sounding thorough than being thorough." The warning adds to a broader industry conversation about AI's impact on junior talent development, following Goldman's own research finding AI's labor market effects remain concentrated among entry-level workers in specific exposed industries.