DeepMind's Hassabis: AI Lowers Trial-and-Error Costs, Could 10x Output of Top Scientists and Creators

Demis Hassabis, co-founder of Google DeepMind and 2024 Nobel Prize laureate in chemistry, said in an interview that AI will sharply lower the cost of trial and error, allowing scientists to test hypotheses faster, creators to try multiple versions, and developers to run prototypes before full production.

He predicted that in the next decade the best scientists and creators could increase their output tenfold. But he also warned that the same tools could generate large amounts of mediocre “AI slop,” making human creativity more valuable rather than less.

Hassabis added that technology companies should not decide the uses of AI on their own, and that education needs to be rethought, shifting from repetitive practice to project-based collaboration. He described the future as “not yet written,” saying the choices people make now will shape where AI goes.