Why spend x amount of years learning it when you can simply ask GPT-10 any mathematical question or have it write any snippet of code?While yes I'm exaggerating, we will always need people who truly understand it to check the AIs output and for novel idea generation (not sure if AI will be capable of this), but at minimum the amount of programmers and mathematicians will be reduced dramatically, and only the absolute elite of the field will be needed.Which brings me to my next question, will all future scientific research just be a matrix of agentic AI researchers computing away in a lab somewhere?If this is the case, it's an ironic twist of fate that the very industries and people responsible for creating AGI / Genius level narrow AI will be rendered obsolete by their creation.
>>16858598AI fails to answer well any question that humans haven't already answered, so it's entirely obsolete.
>>16858598Except a truly intelligent AI couldn't be trusted with anything because it would be built with the personality of the Jewish god, Yahweh. When Jews say they are building a god, it's safe to say that this is what they have in mind.
>>16858643Except for all the solved games, game exploits, new proteins and other discoveries it made before humans did, of course.
>>16858663Name 5
>>16858666CheckersQbertBlock BlasterCoastRunnersTetrisWere all either solved by AI or defeated by novel reward hacking discoveries that lead to previously undiscovered infinite points hack in decades old games.
>>16858598was obsolete before aiany retard can get a drunk whore pregnant, math and coding not required
>>16858643protein folding is the most important solved problem in 100 yearsit was solved by aiyour monkey brain is obsolete
>>16858643>>16858663>>16858669But can AI, for example, provide the missing details in the mathematical proofs of theorems from high level textbooks?
>>16858675Yes it has passed numerous (and even aced some) standardized tests at the doctorate level for a variety of disciplines.
>>16858701Imagine believing those shilled publications. What you don't see are the fifty times it was wrong and needed further prompting to answer correctly.
>>16858674You tell it as if AI look at us: "poor humans can't fold protein, let me help them". And out of generosity solves our century problem in it's free time between suggesting Indian women poop-free cooking receipts and drawing kid-looking porn cartoons characters. All by itself.Maybe you'll appreciate a little the work of scientists that has be done? From knot theory to data science required to setup that narrow-scoped AI, not mentioning all the engineering complexity.
>>16858724>What you don't see are the fifty times it was wrong and needed further prompting to answer correctly.That isn't how standardized tests work or anyone would just get to guess until they get most of the answers right.
>>16858675Can irrational retards, for example, move the goalpost every time their retarded misconceptions are debunked?
>>16858701How can a question about study aids relate to misconceptions. The question was simple, a grad student may be reading algebraic geometry and he finds a theorem with a slick proof. Can Gemini help him break down missing steps in said proof?
>>16858737Meant for:>>16858734
>>16858732Were you dropped on the head as an infant? I don't understand how you can write such an irrelevant comment and miss the point this badly. How standardized tests work is completely irrelevant to how genAI works.
>>16858737>How can a question about study aids relate to misconceptions.The original question had nothing to do with studying, it had to do with producing novel results with the given information and you were under the misconception that AI had never done that before.
>>16858746>How standardized tests work is completely irrelevant to how genAI works.Not when the AI is using its gen capabilities to take standardized tests.
>>16858757Not original as in OP's question but as in the post that was accused of "moving the goalpost":>But can AI, for example, provide the missing details in the mathematical proofs of theorems from high level textbooks?Thread hijacking after worthless OP
>>16858761The post did move the goalposts.If there are documented solutions, then of course AI would be able to fill in any blanks that a random student might have in that solution set, if then entire solution is not documented, then yes it is possible that it could generate what was missing just like it found solutions to games without being given specific instructions.
>>16858598Sometimes AI does mistakes. Reducing the number of paid human vacancies for intellectual work can negatively affect the future of humanity. Watch the movie Idiocracy (2006)
>>16858775Good point, people don't make mistakes, so adding more people to the process can in no way introduce additional mistakes too.
>>16858701>it has passed numerous (and even aced some) standardized tests at the doctorate level for a variety of disciplines.Which conclusively proves that these metrics are invalid, since the latest models still fail at elementary school-level reasoning tasks. But on one hand, this conclusion has been so obvious for the start that the only people needing a proof of it are the ones too delusional to notice when it hits them in the face.
>>16858782But on the other hand*
>>16858780Unlike you and your token-guessing digital analogues, actual humans can do sanity checks on their output, notice their mistakes and even learn from them on the fly.
>>16858757>it had to do with producing novel results with the given information and you were under the misconception that AI had never done that before.A Markov chain can do that. It can't do it reliably, however. Same goes for your imaginary digital friends.
>>16858598>Will 60s sci-fi be real in Two More Weeks?Probably not.
>>16858758You're too retarded to discuss this with. Clinically braindead.
>>16858782{23,3,2} is what Bob was thinking. This type of reasoning is what the chatbot cannot do.
>>16858791A chatbot can't do any kind of reasoning at all. This "problem" essentially spells out its own solution and requires barely any reasoning at all but the wording biases the token guesser towards nonexistent gotchas.
>>16858663>>16858669>>16858674Long computations and sifting through large datasets don't constitute original thought. All the interesting work in these cases was done by people, and AI (rather computers) were utilized to do busywork.>>16858675No, because it hallucinates, cites the wrong people/articles, tries to use claims which do not help in the proof, and even when the proof is correct, it has no elegance and does more to obfuscate than to explain.
>>16858663>Except for all the solved games, game exploits, new proteins and other discoveries it made before humans did, of course.Name a general purpose "intelligent" model that does this in your next post. Notice your blood pressure rising as you begin to realize your error.
>>16858761Sorry, fell asleep. There's two aspects to the post : Novel idea generation and programming / mathematical skills that surpass humans.Even if AI doesn't get to a point where it can generate novel information, it being able to surpass humans in fields like math and programming with known data will still render 99% of the workforce obsolete. Sure, it wont be the cause of innovation and new groundbreaking discoveries, that will be left to the select few who qualify, but it will be doing the busywork at optimality which is what most programmers are doing.That said, I don't see how AI doesn't eventually get to a point where it can generate novel ideas, it having access to nearly all known knowledge in perfect memory makes synthesizing said information to form new discoveries a near inevitability. It can recognize and synthesize parallels between vast and distinct domains, something a human with limited knowledge and memory can't do.I don't think LLMs alone will be the architecture to completely bring this scenario to fruition, could be a combination of ML mechanics and architectures, but I don't see it not happening.
>>16859038Doesn't need to be GP to eliminate the majority of scientist, a superhuman narrow ai specializing in a science field will be sufficient enough for that.
>>1685878760s sci-fi? The gov just launched project genesis, which is literally about running ai agents in robotic labs to automate (and hopefully innovate) scientific research.China has been doing this, just look at the arxiv preprints.
>>16858598Do you not realize that asking the right questions is a skill that is impossibly delegateable to AI?
>>16858598Yes. AI replaces cognition and makes humans far superior than to humans without AI. AI is the future.
>>16858598the reason AI can do so well in math and coding is because these are explicitly discursive areas of knowledge. Earlier bots couldn't do general integration as easy as AI because solving an integral is basically a kind of argument. which is based on language, which AI can imitate very well. AI won't replace fully mathematics because it can for sure reach an answer or conclusion, it doesn't realize what it might have done wrong unless you fact-check it explicitly.
>>16858598>get answer>have no means to verify whether it is correct or not>just have to take it on faith>mfw realizing LLMs are becoming the new religion
AI can only do statistical tasks, it cannot synthesize novel thoughts.>>16858663>>16858669>>16858674These are all statistical tasks.
>>16859355Praise the Machine God!
>>16859275>immediately devolves into irrelevant replacement fetishismConcession accepted. "AI" can't do any of the things listed. Specialized, single-purpose models can. You might as well tell me how smart your GPU is based on how many computations it can do per second. After all, scaling up human GENERAL intelligence until it performs equally at that task would result in omniscient gods. AItards are truly mindless.
>>16859280>the government just launched project Flatten The CurveExactly two more weeks, imbecile. In the meanwhile, you get to fund their ML panopticon project either directly through taxes or indirectly through inflation.
>>16858598would you not like to know more about programing and math to be better at using those services at the very least.OP.
>>16858674>>16859875but , who I am kidding this is the true you OP. you know OP , maybe if you like tecnology and math so much , that you think it would reach a level of secular divinity. you should stop looking at it like a fucking indian and actually appreciate the math and programing that goes behind it.otherwise you risk falling of behind.
>>16859878OP here, didn't write that response about protein folding.Anyways, I do appreciate the tek and mafs, not sure how that's relevant to my OP.
>>16858674based, midwits can only cope against this fact
