Google DeepMind boss rejects AI layoff narrative, urges expansion over reduction
The chief executive of Google DeepMind has criticised the use of artificial intelligence to justify staff reductions, suggesting such narratives may serve ulterior motives such as capital raising.
Demis Hassabis, chief executive of Google DeepMind, has publicly challenged the industry trend of using artificial intelligence as a justification for workforce reductions. Speaking ahead of Google’s I/O developer conference, Hassabis argued that productivity gains from advanced models, such as Gemini 3.5 Flash, should be leveraged to expand organisational output and undertake more ambitious projects, rather than displacing staff.
Hassabis expressed scepticism regarding the certainty with which some industry leaders predict AI will eliminate software developer roles. He suggested that companies blaming AI for recent layoffs may have ulterior motives, such as raising capital, rather than a genuine assessment of technological capability. He described the narrative of job elimination as a lack of imagination and understanding of the technology's potential.
The Google DeepMind chief emphasised that increased engineering efficiency should enable teams to do more work, citing potential applications in drug discovery and game design. “If engineers are becoming three or four times more productive, then we just want to do three or four times more stuff,” Hassabis told WIRED. He noted that while AI models could theoretically rewrite their own code, he doubts this will immediately lead to superhuman-level AI, citing the need for deeper physical world understanding.
Google revealed new AI capabilities at its I/O event, including Gemini 3.5 Flash, a model trained for complex agentic coding tasks such as translating code bases, fixing bugs, and writing operating systems. The company introduced Antigravity, a coding tool offering frontier coding and reasoning capabilities, which Google claims is faster and cheaper than competitors' offerings. Gemini 3.5 Pro, a more powerful version of the flagship model, is scheduled to debut next month.
Other demos included an Android version with built-in AI agents and a refreshed Google Search capable of generating sites or apps on the fly using agentic coding. Hassabis observed that AI has yet to produce a blockbuster app or video game without human assistance, suggesting something is currently missing in current models. He highlighted that increased engineering efficiency allows for more ambitious projects, ranging from drug discovery to game design, rather than job elimination.
