@creator

Technically it has :

nytimes.com

Still probably needs to be more throughly verified before resolving though.

nytimes.com

@Velaris Wow, just found out about this. Yeah, I will probably wait for more verification before resolution.

@Velaris “There should have been no human assistance whatsoever in solving the problem.”

@capybara what are you considering human assistance?

@Velaris @capybara I was thinking about human assistance in terms of making substantial mathematics contributions. Prompting AI or designing a system for AI to solve a problem doesn't count for this. Obviously the AI cannot autonomously solve such a problem without any prompting whatsoever.

i guess there is a grey area in terms of what the prompts are.

@capybara “no human assistance whatsoever” is really strong language and I feel like a mathematician prompting a model should surely meet the threshold of being human assistance, as I doubt they simply said: solve the NS problem.

@waitblock any thoughts with respect to OpenAIs post? https://openai.com/index/navier-stokes-solution/

A few interesting points such as the fact that a group of agents of one model were directed to solve a subproblem, and a further trained newer model was then told (by humans) to start from the proof of that subproblem. All of this was done with agent swarms, and humans ran other agents to summarise findings of various agent groups.

@capybara Based on what OpenAI has said, I'm inclined to resolve this YES if the proof is verified to be correct.

@waitblock have you read their post? https://openai.com/index/navier-stokes-solution/

I’m surprised that the manual coordination and guidance, like “start from the theorem proved by these other models”, doesn’t meet your threshold of your criteria “There should have been no human assistance whatsoever”. If in doubt, I’d suggest waiting until more information is available.

Across most of these markets, the case for No is that humans+AI are likely to be faster to the solution than AI acting without guidance. I feel like this argument holds in this market, and a No would suggest that the real market intent is for “will there exist an AI capable of solving a millennium problem by 2030”. Where solving doesn’t imply first to solve.

Edit: ChatGPT’s own opinion on the resolution: https://chatgpt.com/s/t_6aa708cd25f08191b079e64287892e08

Check out this chat

Here's a chat someone thought you'd want to see.

google deepmind alphaevolve 2 is cooking. trust the plan

I strongly suspect that any such problem solved will be with humans first. It's something people try with each new model as a collaboration. The hybrid will solve any reachable problem before an AI gets the chance to try alone.