Benchmark on AI Science Report https://aiscience.uk/tags/benchmark/ Recent content in Benchmark on AI Science Report Hugo en-us Sat, 11 Jul 2026 00:00:00 +0800 AI Can Write Research Papers. It Can't Tell You Where the Ideas Came From. https://aiscience.uk/posts/ai-scientific-lineage-ideagene-benchmark/ Sat, 11 Jul 2026 00:00:00 +0800 https://aiscience.uk/posts/ai-scientific-lineage-ideagene-benchmark/ <p>A team of graduate students at Shanghai Jiao Tong University recently asked 14 different AI systems a simple question: given two scientific papers, can you tell which one built on the other? The best AI got it right 27.3% of the time. A coin flip would have done about as well.</p>