161–170 of 862 posts
Earlier quoted context omitted.
Wow this is just bullying, it's insane that we are letting these people be in charge of the transition
So someone can make public accusations of theft against you and if you publicly reply denying it you’re the bully?
Can somebody explain: do singularities / blow-ups in solutions have any relation to physical phenomena in fluid dynamics or are they purely artifacts of how the N-S equations may not accurately describe what actually happens in the physical world?
i always found it fascinating how existence and uniqueness of solutions for the basic types of PDEs (Laplace, wave, heat...) follows from boundary conditions of just the right type intuition tells us, i.e. either value or derivative for Laplace (corresponding to fixing voltage or charge on the conductors), both value and derivative for wave (corresponding to initial position and velocity of the parts of the string, as we'd expect from classical mechanics), and also something about the solutions for the heat equation being unstable for negative times (which totally makes sense when you think of "diffusion" -- can't unmix it).
Drama/accusation summary: - Aug 15th: Tristan Buckmaster & Levent Alpöge make progress on a few important math problems, "finite-time blowup with smooth forcing for incompressible porous media, for Boussinesq, and for 3d incompressible Euler." - they do NOT have a proof for the $1,000,000 Millenium Prize problem. BUT, they do claim to have a proof for a similar (non-Millenium) Navier Stokes problem that could help le…
So, leaving aside the idea that OA might've used data from the researchers Codex sessions: Do I understand correctly that the internal OpenAI work on the problems was probably started after they heard Alpoge and Buckmaster had made process by using their models? And they used the publicly available info about the researchers past work to prompt their models? If compute is cheap, and the difficult thing with scientifi…
For math, a field that is built on incremental research it feels like AI labs will do nothing but discourage publishing research at all for fear that they will be able to spend the money for compute that publicly funded academia simply cannot afford.
It feels like publishing anything at this point just means that your work will be fed to a machine that will make sure your work will never been seen by anyone else because it will always be the ones making the “true advancements”.
Perhaps I’d feel better about this if AI labs really existed for humanity’s benefit, but for some reason I don’t think that comes up in their investor slide decks.
Earlier quoted context omitted.
So someone can make public accusations of theft against you and if you publicly reply denying it you’re the bully?
I would assume they are calling the snark from Noam Brown and Dan Roberts bullying, not the somewhat bland denial.
The accused Sebastian Bubeck has denied the allegations on Twitter[0], and various other OpenAI employees[1,2] seem to be mocking another Anthropic employee voicing support for Levent[3]? Things are getting messy. [0] https://xcancel.com/SebastienBubeck/status/20972141224714323... [1] https://xcancel.com/polynoamial/status/2097215233119211902 [2] https://xcancel.com/danintheory/status/2097214838003138603 [3] https://…
If I publish something, and disclose that I used AI for assistance, do I have to credit everyone who previously used the same AI to try the same problem? Because their prompts inevitably made it to the training data for my prompts?
I recommend Terrence Tao's commentary on such a proof : https://mathstodon.xyz/@tao/117219101339291693 Key quote : "Solving the problem by purely AI-powered methods [would be a] net negative for the progress of mathematics."
We are basically mass manufacturing math. Just like you have just 100 designers for a product selling millions of units, you will now need 100 mathematicians to make millions of advancement. Yes you have factory workers, but if we are being realistic they have negative leverage in the world and the analogue of that is not something most of today's mathematicians would want to do. They would want to be in the 100.
Like Tao says, each advancement is now significantly less useful since it yields fewer usable objects. However, we will get many many advancements. Is the tower made with many worse bricks better or worse than the tower made with a few amazing bricks? Depends on the tower. And time will tell.
For some fields of math and some of it's usecases, economies of scale will be positive ROI overall. In others it won't. But we will know which is which only after it's been fully scaled up, which will take 10-15y in my estimate.
Some feel that in the majority of usecases it is negative ROI, some feel the other way, but that opinion is for practicing mathematicians like Tao to hold. Also, some opinions on either side are held in the context of a particular field or practice, and should not be interpreted generally.
Earlier quoted context omitted.
From Buckmaster's text: The route to the Clay problem through a smooth force, options c and d in Fefferman’s statement of the problem, is the route Luis and Diego opened and the one Levent and I had quietly chosen to attack. Almost nobody else I know of was working on it. It is not the direction one arrives at in a few days by giving a model the problem statement. When I heard “forced,” it was a bright red flag. This…
That's a stretch. The Luis and Diego paper was published in 2023 and is included in every frontier model's training dataset. An AI model could independently choose the same path route as Luis and Diego, without access to Buckmaster and Alpöge’s work. And the article states "an insane amount of compute had been used," which implies OpenAI brute-forced their way to a solution. I.e. they searched for every paper publish…
Sorry for random rant but I don't think these statements help your point
The accusation brings up an interesting point. If I publish something, and disclose that I used AI for assistance, do I have to credit everyone who previously used the same AI to try the same problem? Because their prompts inevitably made it to the training data for my prompts?