Our comment in response to the American Science Acceleration Project (ASAP) request for information draws on ongoing research that explores how AI is transforming everyday work in scientific laboratories — not only by accelerating technical workflows, but by reconfiguring how scientific facts are produced, how expertise is recognized, and how epistemic authority is distributed across people and machines. While AI companies advertise that their technology will accelerate certain kinds of discovery, our findings underscore that this acceleration is neither neutral nor uniform. AI adoption in science creates new forms of dependence, new sites of friction, and new questions about accountability, reproducibility, and the value of human judgment.

While we commend the ASAP initiative for foregrounding the need for new infrastructure, metrics, and institutional models to accelerate science in the public interest, acceleration alone is not a sufficient goal. If the United States is to lead the next era of scientific innovation, it must invest not only in speed but in epistemic integrity, continuous maintenance of data infrastructures, and accountable systems of knowledge production. Our comment highlights how AI mutually shapes scientific workflows and offers recommendations to ensure that acceleration does not come at the cost of validity of scientific claims.