Science's Biggest Gaps Stay Unfilled — AI Could Change That

Think of science as a giant puzzle with some pieces missing. A new study from researchers Jiajie Luo and James Evans looked at millions of research papers and used a math tool called persistent homology — think of it as a way to find holes in a high-dimensional map — to identify these missing pieces. They found two surprising patterns.

First, scientists who fill simple, obvious gaps — like connecting two closely related ideas — get big rewards. Their papers draw attention, citations, and fame. That makes sense: these are the discoveries everyone can see were needed. But here's the twist: these celebrated findings are actually rare and predictable. They're like filling in an easy crossword clue.

Second, the study found that the number of complex, higher-order holes — gaps connecting many different ideas across fields — is exploding. Yet almost none of these get filled. Why? Because they're hard to even see, let alone solve. As knowledge grows, these blind spots multiply faster than scientists can tackle them.

The authors suggest this is where AI could step in. Modern AI systems can process enormous maps of data and spot patterns humans miss. If we can teach machines to find these invisible gaps, they might point scientists toward the most promising unexplored territories. This could mean faster cures, new technologies, and answers to problems we don't even know exist yet.