Zeynep Tufekci in the NYT: AI-assisted science may slow science by outrunning expert verification
The New York Times (Zeynep Tufekci)·low signal
Tufekci's September 22 guest essay argues that opaque systems now produce results faster than domain experts can check them, which creates verification, attribution and research-integrity problems. She cites the disputes over recent AI-assisted math work and concerns from mathematicians including Terence Tao. In posts the same day she said LLMs could slow science now and also erode the processes that would produce better science later. This is single-sourced through digest coverage because the NYT page could not be fetched.