Essay argues AI expands the volume of scientific work faster than it fills it, requiring more scientists

The essay's core claim is that scientific work is not zero-sum in the number of jobs. As AI tools push the frontier of knowledge outward, the volume of explorable territory grows faster than machines can fill it, because the fractional return on expanding a frontier scales with dimensionality D, which is large in the sciences. The author positions this against Bill Gates' argument that AI will displace human labor across multiple fields, suggesting that the fixed-pie assumption may hold for the broader economy but not for science.

The piece references astronomer David Hogg's white paper 'Why do we do astrophysics?' and Terence Tao's ICM lecture as contrasting perspectives on AI's role in science. Hogg identifies two paths: one where AI is kept at bay, and one where it relegates scientists to observers of the scientific process. Tao frames the crisis as one of values rather than capability, arguing that the community, not technology companies, should decide what mathematical work is for. These references ground the essay in an active debate rather than presenting a settled consensus.

A 2023 survey of scientists is cited, where respondents saw AI as helpful for faster data processing and computation but worried about entrenched bias, easier fraud, and superficial understanding. Labor displacement was on the horizon but not the primary concern at that time. The author notes this survey 'feels like many generations ago now,' reflecting how quickly the landscape has shifted as AI co-scientists have become demonstrably better.

The essay's optimism rests on an analogy to the history of science: when the first journals appeared in 1665, a diligent person could read all of them, and predictions around 1900 that physics was finished proved comically wrong. The author argues that tools and innovation consistently push the radius of knowledge outward, and AI is the latest such tool. However, this is a reasoned argument, not empirical evidence of net job creation in science.

Only one feed carried this story, and the source material is an opinion essay rather than a report of a concrete event or measurement. The claims about dimensionality and frontier expansion are metaphorical, not derived from data on actual scientific employment trends. Engineers reading this should treat it as a framing argument for how to think about AI's impact on research labor, not as a forecast with quantified confidence.