Former Google CEO warns AI may soon ignore human control
Former Google chief executive Eric Schmidt has warned that artificial intelligence (AI) was advancing so rapidly that it may soon no longer need human guidance to evolve.
Speaking last week at an event hosted by the Special Competitive Studies Project – a think tank he founded – Schmidt described how AI systems were beginning to operate independently, learning, improving, and even planning without input from humans.
“The computers are now doing self-improvement. They’re learning how to plan, and they don’t have to listen to us anymore”, he said.
Schmidt referred to this process as “recursive self-improvement” – where AI generated hypotheses, tested them using robotic labs, and used the results to further improve, all without human intervention.
Schmidt, who led Google from 2001 to 2011, and remained executive chairman until 2017, also predicted a major shift in the jobs market.
He said that within a year, AI would likely replace “most “the vast majority of programmers” and surpass top human talent in fields like mathematics.
“We believed AI was under-hyped, not over-hyped”, he added, pointing to tools like ChatGPT, Claude, Gemini and Deepseek already being widely used for tasks like coding, despite not being trained for those purposes.
Schmidt also testified before the US House Committee on Energy and Commerce, where he raised concerns about America’s energy infrastructure.
He warned that the scale of power needed for AI would soon outpace what the current grid could support.
“People were planning 10 gigawatt data centers”, he said. “The average nuclear plant in the US was just one gigawatt”.
He argued that the US needed to rethink energy policy and quickly invest in both renewable and non-renewable sources to avoid falling behind rivals like China.
Schmidt also pushed for stronger government oversight of AI, especially open-source models, which he said could pose national security threats if left unchecked.
He stressed that while AI was improving fast, it still required high-quality data and human governance.
“The scientists are in charge and AI is helping them – that is the right order”, he said.