AI solves a couple erdos problems and mathematicians are shitting themselves calling for it to become banned.https://siliconreckoner.substack.com/p/the-leiden-declaration-on-artificial
>>108967754AIGODS killed >codetrans>arttransAnd now>mathtransWe vibeGODS are very powerful, we literally can’t stop winning
>>108967754They're not even remotely asking for it to be banned, maybe you can have AI sum the article up for you since you clearly didn't read it.
>>108967766nah the aijeets already used up all tokens to generate tranny porn.
oh no, they'll actually have to catalogue and review papers by writing tests that show the papers work.
>>108967754History fucking repeats.
>>108967924wrong
>>108967924They were right, my cousin was allowed to use calculators throughout highschool and he can't even do a basic ass 2 digit sum without a calculator nowadays.
>>108967972Ok and? The knowledge for math will always exist.Why do you Luddites pretend Computers got rid of books?
>>108967951>Defending copyright.Sorry but copyright is a pro-capitalist agenda.Every human should be allowed to use any idea in their head.
>>108968012The teachers were right, he used a calculator and failed to learn the concepts. Same shit that's happening today with kids, AI does their homework and they don't learn shit.
>>108967754>be ai>have forever memory if you got the storage space>can look up word for word terminology and documentation>can into logicI don't bro maybe they should ban ai, if ai can replace people logic, people literally become robots because they no longer seek independent logical actions.Do you want future society to basically be robot or serfs
>>108968101they seem to think they will personally profit off of machine learning algorithms instead of being locked out by the elites and turned into serfs
>>108967754That’s not what it says though you fucking retard
>>108967951The only one of these that matters is unreliable results. And if mathematics is so great then an error shouldn't be fucking invisible.
>>108968096You're lying anon. The calculator did not make that Leapfrog book disappear. If he didn't learn then that's a CHOICE.
>>108967760>wespeak it for yourselfalso i don't care your fantasies.
>>108967754if you actually read it they;re mad for the same old "we didn't give you permissions to train!!!" thing, not about ai being used in math
>>108967972>>108967924the flip side of this is that American high schoolers are the only fucking high schoolers in the world who are not taught how to use calculators with CAS, so they are unemployable for any work that requires algebra until they go to university
>>108967951only #3 is new from before AI
>>108968730are you saying you cant do algebra without a calculator?
>>108967754it didnt solve the problem it only showed that the problem statement was wrong and introduced a new lower limit. Thats not helpful also reading the proof is very annoying try it yourself no human would write that crap.
>>108968022>copyright is a pro-capitalist agendaCopyright is a mercantilist relic from medieval England, which has nothing to do with capitalism whatsoever, it completely predates all of the schools of capitalist thought.
>>108968216almost every course in my math undergrad doesnt allow calculators its just americans being forced to pay $200 for some shitass TI calculator because of lobby
>>108968743No, I'm saying that in the real world nobody cares if you can do algebra with pen and paper, you're going to be given and going to be expected to use the latest tools that make you the most productive, which today means you have to know how to use CAS software, which Americans are inexplicably never taught about until university.
>>108968795>>108968798so is america bad because calculators or no calculators
>>108968798i bet all the other corporate drones are jealous of you
>>108968211Unreliable results become invisible when each mathematician is expected to review 20 papers a day to keep peer review schedules to 6 months between submission and decision. Remember that review is free labor, nobody gets paid for it and nobody gets paid to publish (people who want to publish have to pay north of $5000 for the privilege).That aside however, mathfags have often been accepting obviously wrong proofs because they were 'dazzled by math', which is kinda points 3 and 4, where now anyone who says 'proof with ai' automatically gets their paper accepted, even with false results, because journals want to publish things that will have a lot of eyeballs on them (that's how they make money). This has been a problem in all fields of science for a long time (in fact, in AI research, it used to be a meme that if you want to get your paper accepted, instead of having good or interesting results, you should just put some bullshit math that isn't even valid and you'll automatically be accepted as having "amazingly insightful grand novel results". We called that "dazzled by math").Result overhyping is a problem because it causes funding funnels into irrelevant areas. Same thing that has happened to structure prediction in bio and general AI x bio bullshit trends where they only fund pure-tech labs with only overfit/fake results to show while serious outlets get nothing because 'eww stinky biologists, tech is the new thing'. Or even in AI research: the only thing that gets funded is muh transformers even though research in other areas like state space models have shown far more promise so far.
>>108967924They're right. They were not protesting against all calculator use, just students using them too early. Kids shouldn't be using AI until college.
>>108967924Damn vibeGODS are fucking stupid. Might as well just suck clammy Sam's cock for a living
>>108967951only points 1 and 4 are validthe rest are just "Muh Copyright", "Muh Inequality", and "Muh Autonomy" which quite frankly no one apart from these faggots cares about
>>108970510Hello chang
>>108970522Ni hao waiguoren
>1. A publication system overburdened with unreliable results>Mathematics is built on rigorous proofs that provide clear understanding. However, AI can produce ‘proofs’ that look convincing but contain almost invisible errors.build better verifiers then, m88. and this is not a real problem - any result will be scrutinised if it's actually useful.>2. Lack of proper attribution and violation of copyright>AI models produce results without citing the human work they build on. This raises questions about recognition and intellectual property.not true. @codex, cite your sources>3. Dependence on AI for results to be considered significant>There is a risk that mathematicians will soon depend on access to the latest proprietary AI technology and expensive computational resources in order to produce competitive results. This leads to inequality between researchers.they should all join team altman and solve the rest of math and we can be done>4. Overhyping of results>Mathematics values work based on depth, difficulty, and significance. Press releases and blogs often make AI claims without scientific scrutiny. This leads to overestimation of AI capabilities and underestimation of human contributions.sour grapes, nigga. two independent solutions to erdos 90 in a week, you niggas have had 80 years. suck it up, there's a new king in town.>5. Loss of autonomy>When technical feasibility or commercial interests shape research, mathematics risks losing autonomy in setting its research agenda.how is this even a new problem nigga. research money always goes to places where there's commercial interest.
>>1089706081. There are no verifiers, you have to go by hand and check the proof works or not. It can take several months to verify (fulltime) for more complex proofs.2. Yes true. Telling AI to cite sources doesn't mean it will. Hallucinations are a major concern.3. The problem is that unless they say "with AI" in the title, their papers don't get looked at anymore, even though it has no relevance at all to whether something is solved.4. Solving things nobody has ever looked at because it's literally useless using trivial proofs that were already used a billion times before is the most retarded cope ever. Most mathematicians have never even heard of erdos before this meme.5. Not true, while it is biased toward commercial interest, that is usually more like hype-based. Example: there is no real commercial interest in pre-llm deep learning or in quantum etc., but those areas have usually been crazy well funded. Furthermore, the funding in pre-llm deep learning was very broad, not single-minded bullshit like we have now. The same is applicable to math for example.
>>108967924They are right.Only a 3rd-world shithole would allow elementary school kids to use a calculator.
>>108967754Why does it seems like codechads are the only people embracing AI?
>>108967754>calling for it to become banned.You must be a schizo, because I just read the article and it says this nowhere.
>>108967754AI democratizes intelligence. It puts retards and pros on the same level.That's why retards love AI and pros don't.Do you like AI? You're a retard.
>>108970698I absolutely believe it as this has also been my experience irl.
>>1089706371. they should use llms, it helps speed things up. they're pretty good at math now2. not a real problem.3. try solving shit that actually matters then.4. yeah man, sure - you'll be saying this shit after they solve navier-stokes in a few months5. >there is no real commercial interest in pre-llm deep learning or in quantum etc., but those areas have usually been crazy well funded.lmao, come on, nigga
>>1089707201. They're garbage at math and hallucinate half the steps, thus defeating the whole point of checking the proof2. I accept your surrender3. Try telling that to the AI tards solving erdos "problems"4. 2 more weeks am i rite? 6 months until 90% of code is written by AI?5. I accept your surrender
>>108968211>And if mathematics is so great then an error shouldn't be fucking invisible.What a stupid fucking thing to say. Let's say AI shits out 1,000,000 billion lines of code, which it can do before I finish typing this sentence. There's a logical error in it, find it. Oh wait, by the time you even reply to this post, it would have shit out 1,000,000 billion more lines of code. Then another. Then another. then another. They don't want their field becoming the indian jeet dumping ground that CS has turned into and that is laudable.
>>108970745>6 months until 90% of code is written by AI?
>>108970757Just one more datacenter! GPT 5 will be AGI, trust me bro! GPT 2 is too dangerous to release!
>>108970767i accept your surrender
>>108970757All that LLM code and yet everything is just as shitty as it was before if not worse, I have yet to see any company improve their software by the 5x productivity boost LLMs supposedly give. Well I guess 200k todo and calorie tracker apps have been "written" with it so from that angle there is definitely a 200x productivity boost to useless garbage.>>108967754Until an LLM solves a millenium problem they are objectively inferior to human mathematicians and potentially harmful as everyone instead of using their own brain just burns 4 trillion tokens hoping something useful drops out at the end.
>>108970773Still waiting for the 90% code written by AI. Where is it, pajeet?
>>108970820nta but if you take into accound how some jeets vibecode 300k LoC simple blog sites I wouldn't be surprised if by sheer combined volume of slop it hits 90%
>>108970864That's a good point, but I don't think that was the context of the quote. The implication was that if you go to a random company, say Google or Amazon, and you check what the approved PRs look like, 90% would be fully AI-generated. Despite how happy the schadenfreude would make me if it was true given recent constant service disruptions, unfortunately I know it's really just jeets behind this so far.
>>108970820if you only look at the vanilla configurations of claude code that mark themselves as contributors on github, those alone added 3 billion lines of code to public repos on github in the last 7 days>>108970876google claims over 75% of all new code being ai generated and they have the worst coding modelsyou're simply detached from realityllms completely changed swe over the last 6 months and they will do that to everything else over timeyou will be worse at the thing you love than a machineit's time to grow up
>>108970898Not sure if retarded or merely pretending.jpgNot a good look for aitards as usual.
>>108968739Amplification of existing problems is a valid concern.
>>108968730>>108968798In Russia, nobody teaches you how to use CAS in high school either. But if a kid isn't completely brainlet tier, they’ll find Wolfram Alpha on their own and figure it out. It’s not rocket science. I literally learned how to use Wolfram at a basic level just to double-check my work back when I was in 8th grade (I was around 14 year old at the time).We aren't allowed to use calculators on math exams at all - not even basic-tier ones. I don't see the point in teaching high schoolers how to use CAS. Anyone who actually needs it will either teach themselves or learn it in uni. High school math should focus on teaching functions, graphs, trig, logs, problems with parameters, plane geometry, and solid geometry. Plus, covering limits, derivatives, and basic integrals (which are part of calculus-1 in most universities too).You don't need CAS for any of that, honestly, it just rots your brain.
>>108971389I think you could spend a few lessons here and there using CAS to explore those things though.
>>108967754Let me get this straight >Woah we got these super hard math problems no one can solve >Will we ever get these proof solved???>Enter 16 year old Aiden with Claude>Solves proofs >NOOOOOOO WE MUAT BAN AI AHHHHHHHHHHHHHumanity can't get wiped out fast enough
>>108967951>2+2=4™
>>108971442There's nothing surprising about status quo enjoyers wanting to shun disruptors to said status quo. If your livelihood was completely dependent on processes being slow, you'd want them the stay slow too.
>>108971442Not even remotely. Stop getting your news from hallucinate.ai, get it instead from actually looking it up.
>>108971442>Enter 16 year old Aiden with Claude>Spams bullshit nobody can prove >Please stop>NOOOOOOO FUCKING SNAILCATS I SHOULD BE FLOODING YOUR FIELD WITH PROOFS BECAUSE.. I JUST CAN OKAY? GIVE ME MY GRANTS AI bubble can't pop soon enough
>>108968798>>108968730>No, I'm saying that in the real world nobody cares if you can do algebra with pen and paper, Everyone cares, because algebra is about representing the real world with numbers and variables and not solving textbook problems with pen and paper OR CAS, you unemployable jeet brainlet.
>>108967754>AI solvesIt didn't solve shit jeetbro.And if LLMs prove something and it is checked with an LLM then it's not a valid proof.
>>108967760>we>owns no hardwareDon't speak for your betters you piece of shit.
It makes sense.LLMs are designed to produce something that looks like something that is correct, not something necessarily correct.So now they spit out large complex proofs that look correct so even their mistakes will look like something correct, creating overwhelming work for mathematicians trying to verify them.
>>108971647People are designed to produce things that they think are correct, not things that are necessarily correct. People also sometimes just lie and try and pretend that things they know are incorrect are correct.This isn't anything specific to AI.
>>108971669That is true. The difference is that here is intentionality with people, not with the "AI". This means that if someone submits a proof, they either think it's correct or can trick someone into thinking so. For the AI, it makes no difference. This is why the flood is a problem. Not to mention the lack of accountability issue.
>>108971669>People are designed to produce things that they think are correctwatis that something from the Hindu faith or something?
>>108971462More like we don't want it dependent on a couple of rich assholes who own all the compute
>>108968101Humans don't operate on pure logic like computers. We are more like a governed interiority that actively manages what is admitted into the conscious whole. An AI has no governed interiority bounded by its thermodynamic processes. That is why it can hold so much logical data within its latent manifold and yet cannot be the one to say, "There is something undiscovered here that is worth investigating." It always requires a human's input first - a governed interiority that decides something is worthy of attention.
>>108971836>>>/x/Never come back
>>108971854>philosophy 101>/x/ rather than /lit/Do you put ethics and UAP sightings in the same cognitive bucket too?
>>108971872>sub-0 IQ retardo thinks his profound brain problems are philosophylmao
>>108971872>sub-0 IQ retardo thinks UAP sightings don't carry heavy ethical implications that are worth figuring out
>>108971899Holding ethics and UAP sightings in the same cognitive bucket is how we get people amending their Golden Rule into:>Treat others how I want to be treated (this doesn't apply to suspected aliens in human disguise)
Good. The ultimate mission of humanity should be to leave the planet as soon as possible and explore the planets. AIGODS are winning desu. Maybe the elites can actually focus on that when labor is solved.
