Talking in a recent interview on artificial intelligence (AI), IBM Vice Chairman Gary Cohn dismissed fears that AI will devastate employment, instead positioning the technology as the next step in workplace evolution that will ultimately create more opportunities.
“I’ll give you the rest of the time I’m up here to come up with a major technological advancement that destroyed jobs,” Cohn challenged, drawing parallels to historical innovations like the internal combustion engine. “We needed mechanics, we needed gas stations… there were more jobs created because people could go 50 miles a day or 100 miles a day versus two miles a day.”
The former Goldman Sachs president and Trump administration economic adviser believes AI will follow this pattern, shifting workers from unfulfilling roles to more satisfying ones while boosting overall productivity.
Cohn explained that IBM has already implemented an HR chatbot that provides employees with instant information about benefits, compensation, and documentation requests. This automation allowed the company to redeploy approximately 600 HR staff to more valuable positions while simultaneously improving employee satisfaction through faster service.
Regarding the future direction of AI development, Cohn said the efficiency of targeted smaller models over massive general-purpose systems.
“The small models are much more efficient, they compute faster, they take less power,” he noted. “I think most companies will engage with as many small models as they need to get done what they need to get done.”
While acknowledging that large language models will continue to exist as “catchall models for everything,” Cohn sees specialized AI as the more practical approach for most business applications.
Looking beyond AI, Cohn predicts quantum computing will arrive sooner than many expect — potentially within five years — driven by significant private capital investment and rapid advances in error correction capabilities.
When addressing regulatory concerns, Cohn advocated for a balanced approach: AI applications in already regulated industries should face appropriate oversight, while more casual uses likely require minimal intervention.
