Goldman Sachs exec warns AI efficiency may come at cost of employee reasoning skills
A Goldman Sachs partner is sounding the alarm about artificial intelligence eroding one of the most fundamental human capacities: the ability to reason.
New York-based Chris Churchman warned in the bank's "Exchanges" podcast series that widespread AI adoption risks hollowing out employees' cognitive capabilities over time.
"I think 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. I think there's a real danger of that," Churchman said.
He drew on historical precedent to illustrate his point — noting how humans lost wayfinding instincts when navigation tools emerged, and how the printing press made memorization obsolete. With AI, he argued, the faculty now at risk is reasoning itself. His prescription was deliberate: employers must make protecting human thinking a conscious priority.
"My North Star here is I feel a great responsibility building products for the firm such that in 10 years’ time, a new joiner to the firm, their friends say to them, 'Oh my goodness you can handle that much reasoning?'"
This outlook should be of concern to many in HR who are rolling out training and policies to encourage the greater use of AI in the workplace. While the short term may produce results, the long term could lead to major challenges in employee skills.
The illusion of perfection
Cognitive offloading has long been part of our daily lives — automated calendars to remind us of appointments, calculators to handle large numbers, according to Fuat Ramazanov, an AI researcher and member of the Postplagiarism Research Lab at the University of Calgary.
While AI may enable employees to finish a job faster, is it the best output? Instead of defaulting to speed as the primary KPI, managers may want to measure new skills development or critical analysis, he says.
"Young people may not have all the skills to critically evaluate AI-suggested ideas," says Ramazanov. "They may perceive them as interesting and worth exploring, whereas more experienced people engage more deeply and look critically."
He points to AI's illusion of perfection as the central problem. Because AI output can be nearly perfect with little effort required, employees are robbed of the productive struggle that builds real capability.
"Managers are now expecting people to work faster, but it's important to understand that creativity requires learning from mistakes," Ramazanov says. "We have to sometimes prototype something, which may not be perfectly usable, and then continue evolving until we get to the final product."
That productive struggle, he says, is where creative thinking develops — through two distinct cognitive processes: divergent thinking, which is identifying possibilities, and convergent thinking, which is making decisions about those ideas. Generative AI is excellent at the first and poor at the second. It can flood a room with ideas, but it cannot meaningfully evaluate them.
"Individually, creativity is enhanced, but collectively, originality is reduced," says Ramazanov, who is also director of the International Business Management Program at the Ascenda School of Management in Vancouver.
Getting AI integration right
The fix, Ramazanov argues, isn't to pull back from AI — it's to be deliberate about how humans engage with it. He says employers' number one priority should be training people not just on the technology itself, but on metacognition — understanding how humans think, including our own limitations, and how our existing skills can work in tandem with AI.
This is particularly important for younger employees who may be AI natives but don't yet have the practical experience to properly assess AI results. Managers and senior employees who already have hands-on experience can act as role models for junior staff and walk them through evaluating AI work with a critical lens.
"If you have new employees in the workplace, we really need to help them to understand the nuances and realities of work rather than simply giving them a task at hand and saying, 'Hey, just go ahead and do it,'" says Ramazanov.
He recommends that HR teams conduct job analyses to understand what tasks benefit from AI use versus which ones are better with human control.
"We need to preserve apprenticeship and junior-level work," says Ramazanov. "We need to appreciate the new generation of employees who are stepping into work after college. They need a chance to learn, and learning involves making mistakes; creativity involves making mistakes. I don't say they have to fail, but there should be some task they have to learn by making mistakes in a safe environment."
