Francesco Maggi(@Frances01896069):They have withdrawn 3 connected papers “solving” a case of the Hodge conjecture. One of them had a sign error that invalidated the argument, and with it the other two. The subscription models can easily find that error. So they rushed to put out a partial result on a Millennium problem without even checking it twice. That’s the level of care, the seriousness, with which mathematics was approached here. LLMs are an amazing tool for doing math. But using them like this is just irresponsible. Also, they have already benchmarked these models on proving big theorems. I’d love to hear what investors think about all of this.

2026.10.08 12:07

They have withdrawn 3 connected papers “solving” a case of the Hodge conjecture. One of them had a sign error that invalidated the argument, and with it the other two. The subscription models can easily find that error. So they rushed to put out a partial result on a Millennium problem without even checking it twice. That’s the level of care, the seriousness, with which mathematics was approached here. LLMs are an amazing tool for doing math. But using them like this is just irresponsible. Also, they have already benchmarked these models on proving big theorems. I’d love to hear what investors think about all of this.

108

2.5K

336

コミュニティへ転送

人気のあるユーザー