In November 2020, organizers of a key scientific meeting on protein structure prediction announced that artificial intelligenceArtificial Intelligence (AI) (AI) had solved the protein-coding problem, a scientific challenge which had stood for over fifty years. The scientific and popular news media quickly lauded this breakthrough. Since 2020, however, a less visible (and less celebratory) set of debates has continued about what exactly AlphaFoldAlphaFold has achieved and how important those achievements are for science and medicine. These discussions provide some important clues as to as how the usage of AIArtificial Intelligence (AI) in scienceAI and science may play out. They suggest that AIArtificial Intelligence (AI) will play an important role in redefining what counts as an interesting or important problem; as problems become tractable for AIArtificial Intelligence (AI), scientists are likely to reorient their work towards other questions and problems that they perceive may not be AI-solvable. Moreover, a shift towards AIArtificial Intelligence (AI) would entail a further shift towards data-driven forms of science that have developed clear metrics for assessing solutions; it also may cause scientists to re-imagine and re-configure problems in these terms.