@tonyrigatonee The left's acquiescing to any position is on woke just to get nonwoke left wing economic progress and you respond by making fun of them for reaching out
You actually deserve the most neocon AIPAC first US last RNC nomination for 2028 since you'll deepthroat Israel to own the libs
@ArunAnnow Use cursor the way you would currently use vim to edit text and carefully watch all changes as they happen? This has been trivially a better way of writing code than manually for years now
@rod_mallo@madelinelawren Obviously yes because you have a row and 3 children with bags they can't lift into the overhead unassisted and you're not going to leave them in a busy airport
@ConvexMarketsAI@ArielKwiat when I asked this exact sum plus 2 digits it was 1e-9 incorrect, obviously it's contrived but my point is the transformers themselves are not good at this task
cli agents are just larp, there is no reason whatsoever to not use a gui harness it’s better in every possible way. all of this nonsense is the vim vs vscode nerds of this era
@0xMaxSF@10x_er Honestly if you're not reading a single line of code that you've written then your project has a theoretical maximum size because your prompts will become worse and worse at describing what the actual issues are when you are only describing them from a frontend perspective
gl tho
@ConvexMarketsAI@ArielKwiat this is actually incorrect still for numerical
heavy algebraic is decent and pure is PhD level but if you e.g. ask it to times together 50 random 2 digit numbers it will fail when "pure" (though of course who cares, it's still got capacity to write correct inputs to useful tools)
@ArielKwiat "Hi, computer scientist here! Actually a pure LLM is what thinking *really* is
I say this not from understanding Fodor's three tiered cognitive system, modularity or 5 decades of related philosophy, instead from the fact I can code meaning I have a god complex and am never wrong"
@tonyrigatonee The left's acquiescing to any position is on woke just to get nonwoke left wing economic progress and you respond by making fun of them for reaching out
You actually deserve the most neocon AIPAC first US last RNC nomination for 2028 since you'll deepthroat Israel to own the libs
@zagrebbi The Fermi paradox is that every time the first developed society in the cosmos begins to look at the stars they conclude they are simulated and then give up on acting like the universe conquers that we are!
Seriously. Reality is 14 billion years old and we assume we're not first?
@zagrebbi I'm from ASI utopia! Despite entropy I extremely desire to simulate just before the singularity for some never-discussed reason
Y'all are definitely not coping over living as the only generation who can both enjoy technology and have the climate not be perma fucking disintegrated
@LeusKapus@garvvee Won't somebody think of the quantative trader hiring pipeline?
And what if we meet N dimensional Aliens having trouble putting spheres inside hypercubes 🥲
@stalkermustang Soundness at a kernel level is in the process of being proven for lean 4!
Then once that's completed this issue will permanently not exist anymore (in the sense that future versions of lean can be proven sound if what is written can be automatically converted into lean 4 😄)
@GaryMarcus@CristianFl46703 It's a pure transformer-decoder hooked up to tools. Why is this an issue?
Purely your brain did not write this tweet it was your tools the fingers but casually your brain was responsible, same way AI caused these proofs
If anything, complex tool use is a sign of more intelligence
@GaryMarcus@emollick@polynoamial >erm I see you've automatedly formalised a proof of the Riemann hypothesis in Lean. However this is NOT pure because you were allowed to search the internet, try again and do it yourself
Do you understand how silly this objection is, assuming these results weren't human-in-loop?
@InverseMarcus I guess, but also it probably did, right? E.g. it probably searched the internet for arxiv when researching (which again is not purely transformers)
My broader point is "why does this matter even if it did, because this is an insane metric of ability noone is ever using nowadays"
@psyunodim@polynoamial@OpenAI There could be subtle hallucinations within a paper that experts in the field would need to carefully look at to discover, which is why they've developed a library of maths inside the programming language called Lean which can be used to check whether a proof is correct instantly