AI MAY MAKE SCIENTISTS WORK MORE, NOT BETTER

The study models how time savings from AI tools could backfire. When language models reduce the hours needed for existing research, freed-up time becomes available for new projects rather than refinement. The result: researchers spread themselves thinner across more work. In two of three modeled scenarios, individual publication quality drops. The mechanism is straightforward—as remaining research hours become more valuable, they get reallocated to quantity over depth. This assumes AI tools work flawlessly. Even perfect language models cannot solve the fundamental economics of research time. The findings suggest productivity gains from AI depend heavily on how institutions and researchers choose to spend the hours they gain. The research adds nuance to ongoing debates about AI's role in academia. Time savings alone do not guarantee better outcomes if organizational incentives push toward volume over rigor.