There is no point in doing math anymore. OpenAI solved math. It's over. Leave academia.
womp womp
>>17072817Someone has to verify the leanslop
>>17072820For now.
>>17072822Why didn't OpenAI use their super duper intelligence to clean up the horribly written papers?
>math career
>>17072825If you think that's bad you should see the code they generate. It's no better.
>>17072827absolutely dreadful when I see the intellectual elite pronounce coding and math dead, civilization went very wrong somewhere
>>17072828That's the thing. AI can generate code (and math) in hours it would take entire teams months to come up with. It's churning out so much content it's near-impossible to check it all for correctness or quality, and bosses aren't giving code monkey the time to do so.
>>17072817You mean in Academia or the private workplace/government?
>>17072817academia died the day it became yet another venue for careerism
>>17072836So from the start?
>>17072836So that 1600s then? That shit goes on everywhere, in every job, and always has done.
it doesn't matter if we understand the code as long as it worksit's the same with proofs, they are math codehuman provers are the same as coders. if that's all you can do then your job is fully automated nowif you are already in a phd just chill, let chat do all your work. you basically have a do-nothing job. so i dunno what they are not happy about
if the ai is so smart why can't it present a flawless proof that's instantly convincing to humans
>>17072848The questions it's answering are either hard, complex, or highly technical - or a mixture of all three. That doesn't exactly lead to them having short, concise proofs.
>>17072817>Leave academia.To where? Everything is getting automated and the only way to get a job is being a woman (which I will never be).
>>17072849When exactly in the history of academia has this ever stopped people from presenting understandable proofs?>short, conciseI never said this. I just said one that's convincing to humans. So far human mathematicians find the AI output to be slop that they can't follow at all, not that it's necessarily "too long" or anything.
>>17072896Even a flawless proof can take days or weeks for experts to review. The Fermat's Last Theorem and Perelman papers took months to analyse.
>>17072842And what about when it doesn't work? If we don't understand why it doesn't work how the fuck are we supposed to fix it?
>>17072817>mathNext time learn something that is not a meme
>>17072832ai code isn't maintainable. you have to custom generate it on the fly for that one specific use caseI worry for all the opensource software project getting polluted by slop code because they don't have the resources to audit it.
>>17072817TOTAL MATH DEATH.
>>17072902Give a better prompt to AI. Radical idea, I know. You don't need to know how something works and fails to try something new. So many scientific discoveries are found this way. They try random things without understanding it to see what sticks. After it works consistently for a while, if you're lucky, you can pinpoint how it works.
What makes you think mathematicians will continue giving AI their research at this point?
>>17072848it's not smart it's stealing the research actual mathematicians are feeding into it. And Sam Altman wants to make a living out of this
>>17072842>it doesn't matter if we understand the code as long as it works>It's the same with proofs, they are math code"Green ideas sleep furiously on a painted number seven"What does the above mean? It's the same with Lean, it might make sense or it might not, there might be logical leaps in the code that self contained are true.How do you know if it works.
Mathie are you ok? Are you ok? Are you ok mathie?
>>17072842>it doesn't matter if we understand the code as long as it worksSomehow ritualistic paganism has returned.Make sure to say the right prayer so that the AI gods will deem you worthy of receiving their holy solution.
>>17073111You don't understand how your post appears here and yet you trust that "it just works". Almost all human actions operate on this principle that someone knows how it works, in this case the someone just happens to be AI
>>17073115>Almost all human actions operate on this principle that someone knows how it works, in this case the someone just happens to be AIPlese learn the basic architecture that they use.
>>17073111>Somehow ritualistic paganism has returned.If we're talking about programming specifically, it MIGHT have something to do with the computing ecosystem itself (d)evolving into an incomprehensible, highly layered of pile of trash held together with duct tape and prayers. You can see the cracks anywhere you care to look, but no one's ever gonna look at it and go like: gee, maybe the entire field of computing is built on a fundamentally ill-conceived infrastructure, so let's scrap it, learn from our mistakes and try again. No, it's always gonna be ad hoc patches patching patches. It was basically game over when pic related lost.Maybe some day "Computer Science" will become an actual science, in the sense that you're gonna have people empirically probing and theorizing about the computing ecosystem like it's an alien landscape driven by its own laws, indifferent to any human design principle. But until then, you're gonna have to deal with these retards praying to their false AI gods.Come to think of it, didn't McCarthy invent Lisp specifically for symbolic AI? AI could've been lawful neutral. Instead, it's chaotic evil, just like computing as a whole.
>>17073124>Maybe some day "Computer Science" will become an actual science, in the sense that you're gonna have people empirically probing and theorizing about the computing ecosystem like it's an alien landscape driven by its own laws, indifferent to any human design principle. But until then, you're gonna have to deal with these retards praying to their false AI gods.Like any good ecosystem, this one will be prone to natural disasters and like any good cargo cult, this one is gonna blame an angry god.
