Scientists can save time with AI while accumulating more ideas than they can test. Google's ATLAS research update describes a survey of more than 600 scientists in the United States and United Kingdom who reported saving almost seven hours a week through AI use.
The same research reports time spent validating AI outputs and a backlog of untested hypotheses, with physical experiments and clinical validation emerging as bottlenecks. These self-reported time savings therefore tell us about parts of the research workflow. Turning them into discoveries still depends on the capacity to check ideas and conduct experiments.
