MIT Tech Review AI's analysis correctly identifies a critical bottleneck: the idealized conditions that enabled AlphaFold are vanishingly rare. The industry's current rush to fund 'AI for science' startups often overlooks this fundamental data scarcity, potentially setting up a wave of disillusionment. The pivot toward reasoning agents, as the article outlines, appears more immediately practical, mimicking the messy, iterative process of human research rather than demanding pristine datasets.
However, this shift also suggests a less revolutionary, more incremental acceleration of science, which could recalibrate investor expectations and redefine what constitutes a 'breakthrough' in the coming years.