AI predictions and the expansion of scientific frontiers: Evidence from structural biology

Artificial intelligence holds the potential to expand the frontier of scientific research1, yet recent work has raised concern that it may instead narrow scientific attention to well-established areas2-4. Here, leveraging the 2021 release of AlphaFold25 as a quasi-experimental opportunity, we provide field-level evidence that AI can redirect collective attention toward more novel research targets. Tracking 245,396 experimental structures in the Protein Data Bank6, we show that a long-running decline in the study of novel proteins halted after AlphaFold2s release, with the shift concentrated among studies citing AlphaFold2 and targets with high-confidence predictions. This pattern extends to 248,191 downstream papers that consume structural knowledge, where engagement with genes lacking experimental structures and with understudied human genes increased since 2021. Amid rising concern that AI may reinforce scientific canons7-10, our findings offer an early field-level case where AI predictions expand scientific frontiers, consistent with the idea that the real-world consequences of AI on science depend on where their informational gains are greatest. These results suggest AI can complement human knowledge and redirect collective attention in science, with broad implications for emerging AI for science models.