دیدید ترسم بی‌جا نبود؟
آقای Jukka Suomela توی لینکدین امشب پست گذاشته:
I think our research community should seriously start to have this conversation:

How do we plan ahead for the day OpenAI releases their next-generation model?

That day, dozens of students, postdocs and professors all around the world will feed the same standard open questions to the model. Say, in our case, questions like "what is the exact complexity of distributed (Δ+1)-coloring in the LOCAL model?" They all burn their tokens and all of them get a solution, essentially the same solution. Then what? All of them rush to understand the solution, write it up nicely, and try to be the first to submit somewhere? And the incentives are set so that the one who dares to submit half-baked slop first wins, and dozens of smart people wasted their time and effort?

How could we plan ahead so that the end result is not just tons of wasted effort, time, and money, with huge amounts of redundant uncoordinated parallel work?