They tell you that they used AI to optimize a solution, you think of "simulated annealing" or similar.
But no: you are told LLMs were used to sort through «hundreds of thousands of pages of scientific literature»¹. The big question does not seem to appear tackled in the article: how did they use the unreliable NN when needing to obtain objectively optimal results?
But no: you are told LLMs were used to sort through «hundreds of thousands of pages of scientific literature»¹. The big question does not seem to appear tackled in the article: how did they use the unreliable NN when needing to obtain objectively optimal results?
¹(«Large language models extract chemical compositions and material types of 14,000 materials from 88,000 academic papers» - https://www.nature.com/articles/s43246-025-00820-4)