Evaluating academic publication quality is a slow expert task. Until recently, the main quantitative indicators used to support it were citation-based, such as article or journal citation rates. Now, however, there is evidence that scores from Large Language Models (LLMs) more strongly associate with research quality in most fields. This talk will review the evidence so far, discuss why LLMs might have this capability, and introduce questions about how and whether it is possible to use them to support important research evaluations.

Free access until full capacity is reached.