Here's AI:-it works but it has nothing to do with intelligence-it knows things it has been fed (initial training of AI) but then it sometimes vomit text where information disappears-you can teach a parrot to speak, with infinite memory capacity parrot could recite entire shakespeare, it wont make parrot nothing more than a parrot though-if you give a large language model hands, it wont be building a nuclear reactor even when all nuclear knowledge had been input in it previously-cant you see its a random text generator with infinite text resources-we could have built the same in 1980 if computers back then were on the level they are now (its about CPU power, amount of memory, amount of storage space and speed of storage space, thats why datacenters are now being built, contrentrating billions of computers into the use of "AI" it needs parallel computing)
Well AI right now has 0 physical senses. Imagine what it can do when it can finally see and experience the real world just like us.
>>537702131I like the parot analony
>>537702198The human brain is a stochastic parrot too
>>537702189it needs a different underlying technology to develop similarly as a human being develops his intelligenceand I am not claiming it needs to be biologicalbut here's the thing: a human doesnt learn in the same way current AI does
>>537702379Because humans have 5 different senses to get information from and learn. AI right now has 0 but they are working on the sense of vision.
>AI is a *FINANCIAL* scamftfy OP
>>537702545its not the same at allif we give AI those 5 different senses, it does nothingwhen we input this command into AI:>python train_model.py training_data.csv --max-depth 10 --learning-rate 0.2it will start to use its 5 senses until its commands are deemed completed by the computer interpreter
>>537702545big if true
>>537702131>it works but it has nothing to do with intelligence>>537702131>random text generatorcope>>537702131>parrot>>537702198>I like the parrot analonyAt least complete your semiliteracy and read the whole paper that coined that term.dl.acm.org/doi/abs/10.1145/3442188.3445922Then you can tell us more about stochastic parrots and environmental racism.
AI just helped me fix an administrative fuck up at work, so that's nice
>>537703022its an infinite random text generator and infinite parrotthose create pretty good illusion of intelligence, unlike a real parrot or text generator from year 2006
>>537703607You are missing the point. A "stochastic parrot" is an oxymoron. An LLM is neither random nor does it word for word parrot existing text. You on the other hand are regurgitating woke scientists without showing any sign of thinking for yourself.
>>537703607LLMs have solved a previously unsolved math problem by applying algebraic thinking to geometric problems - a real innovation a mere parrot could never do.You have to concede the parrot analogy or admit that it has zero implications and you're just using it because you think it makes you sound smart.
>>537703607(it doesn't btw)
>>537703803Can any LLM count to 10? 100?
>>537703876
AI can help you avoid household dangers from storing certain chemicals together. Someone could unknowingly cause an explosion or deadly gas. The average woman probably keeps these in their medicine cabinet.
>>537702131Kind of...The whole thing is absolutely retarded and every single problem they have with ai is solved by "train it more", to make an actually smarter ai we need more intelligent training and data processing but its hard so nobody wants to do that. Instead they just get a gorillion dollars from private equity then they flush ut into the nvidia samsara cycle of death, exit and invest their newfound wealth in something real like gold or something
>>537703803>LLMs have solved a previously unsolved math problem by applying algebraic thinking to geometric problemsNo they didnt, that one is a scam and they hope no human being will start to try to find it out, it would take a lot of effort, several years to find it out and they are counting on thatIts a proof that looks like a proof well enough so that it requires immense work to prove it false, but it will be false eventually.
>>537703876can you?
>>537703925software made in 2006 can do this just as well, AI is not an requirement
>>537702131>AI is a scamIt was never meant to be a consumer product. It was always meant to be used to monitor your every movement and action.
>>537703942>Its a proof that looks like a proof well enough so that it requires immense work to prove it falseCope. Mathematical proves can be checked automatically. It just shows you know nothing about the field. You are pathetic.
>>537703912
>>537703963>software made in 2006 can do this just as wellNo it can't. Why are you insisting on being retarded?
>>537704054>AI can help you avoid household dangers from storing certain chemicals together.You are saying software from 2006 cant tell you this? You are the retard here. And obviously you dont know a single thing about compter software and software development.
>>537703949Much better than germans, french or dutch.. >9 over a rowingboat plus 1 benis = ten ten ten tennis
>>537702189> ChatGPT, imagine the smell, then continuously generate an even more intense variation of it
>>537704093Tell us which software in 2006 could do that?
>>537704124>notepad
LLMs just regurgitate whatever the most updooted posts on plebbit have said on topics; of course it can tell you how to change a lightbulb, but for any topics that involve any degree of nuance, never mind specialised topics, they fall apart. I can easily win debates against such things because they parrot midwit stances without actually understanding the implications of those positions. In their defense, actual midwits do this as well. So basically, we invented pocket redditors that help you correct your wrongthink and check your privilege. Great, the world definitely needed more of those.
>>537703607Dude how do you even prove that the human brain is any different? That it's not a stochastic parrot either?
>>537702632Shit is so fucked
>>537704124>Avogadro>ChemBioDraw>HyperChemYou need to input into database what chemicals you have in storage and it produces a warning if it sees there are those which create explosive compounds if mixed together
>>537704234pathetic. You could have at least asked the LLM to pull out some rigid expert system for you.
>>537703022LLM's aren't sending their best. You will never be intelligent (or a woman).
Doctors parrot information
>>537703963AI can help you plan meals, find recipes, and develop a shopping list.
>>537704296I think it is pathetic and inappropriate to use LLMs when having a conversation with another humanbeen. >autocorrect is also humiliating
>>537704265it looks like human brain is fundamentally different because it can learn on its ownAI hasnt been able up to this day, learned things on its own, humans have to input information into it for a long timehuman (or a cat) is autonomous, AI is not, its fully dependent on user
>>537704272>You need to input into database what chemicals you have in storage and it produces a warning if it sees there are those which create explosive compounds if mixed togetherThat's a shit idea and would never work in practice.
>>537704364I am pretty sure someone made a computer software for this kind of thing from 2000 to 2017 when there was not yet a possibility to use AI for suchBut I'm no expert on meals, recipes, shopping lists so I cant tell anything for certain
>>537704455it is probably shit feature but it exist in the software, which was kinda the point
>>537704510ok wow now everybody is convinced AI is a scam.
>>537704443You don't even know what you are talking about. There's self-supervised learning and reinforcement learning now. AI learned to play the game of Go by itself and beat the world champion.
>>537704626you know nothing what YOU are talking aboutgo game AI has nothing to do with LLMs>it was released in 2016, not in 2020s>it doesnt know anything but is good at Go game>its not even LLMGo-AI was made with Reinforcement Learning (RL) network and utilizes algorithm known as Monte Carlo Tree Search (MCTS)BOTH of which are not involved in LLM at all
>>537704443>it looks like human brain is fundamentally different because it can learn on its own>AI hasnt been able up to this day, learned things on its own, humans have to input information into it for a long timeHumans also don't learn from nothing. Everything you see, hear and feel is training data for you.
>>537704784lol your knowledge is outdated. Reinforcement learning and Monte Carlo Tree Search is USED in modern llm reasoning models.
>>537704856>Everything you see, hear and feel is training data for you.AI cant do it on its own. It needs to be told "look at this, hear that, and start learning"Human learns automatically wuthout being told to "start learning now"
>>537702131The real problem with it is that you can't derive world knowledge from next token-prediction alone. It's a guessing machine. It's as when they say the "letter" rather than the "spirit" of the law. It only understands the letter.When people write or read something, that sequence of characters is entangled with some kind of mental content or picture. That content is completely absent from these models.
>>537703984>Cope. Mathematical proves can be checked automatically. It just shows you know nothing about the field. You are pathetic.er ist nicht ich
LLM is good at anything that requires words that includes programming languages. Our entire technology interface is built to work with human inteligible programming languages, so LLMs will get better at that.
>>537702131I feel they are trying to sell the public on so-called AI with shiny use cases that are somewhat unnecessary and not the best use of AI, just so they can farm everyone's data and make everyone dumb and dependent and create better targeted marketing and push out the latest groupthink updates. Meanwhile, the real but mostly boring data-crunching uses will remain largely hidden behind the curtain.I saw an interview yesterday with the VP of Meta wearables, he was asked why people should buy the glasses. His favorite example? He was in a car with his kids looking for a McDonald's, but the glasses knew his tastes, and instead suggested a nearby diner that specialized in burgers and banana milkshakes. Like barf me, I'm supposed to buy your shit to data farm and track me just so you can steer me to some dumb diner for a dumb burger instead of the other dumb burger I was going to eat. Yeah, I need that in my life.
>>537705816>I saw an interview yesterday with the VP of Meta wearables, he was asked why people should buy the glasses. His favorite example? He was in a car with his kids looking for a McDonald's, but the glasses knew his tastes, and instead suggested a nearby diner that specialized in burgers and banana milkshakes. Like barf me, I'm supposed to buy your shit to data farm and track me just so you can steer me to some dumb diner for a dumb burger instead of the other dumb burger I was going to eat. Yeah, I need that in my life.exactly, actually mindblowing use cases are missing..like "is anyone I see on the street right now publicly known as dangerous felon">searching<I cant tell you that
>>537705816>>537706133Imagine my glasses could scan all the women who are ovulating while they walk by.https://youtu.be/A7wS6B-xQas
>>537702131Boy are yoy still seething about ai "stealing" your job? Get over it loser. And stop speaking english you have nothing of value to say.
>>537706660>your native language is russian>AI cannot steal my job because it has no hands and traditional robots havent been made to replace me for some reason (too expensive?)
>>537706133>Real life scanning with AIyeah, that would be one actual use case, basically the visor from metroid. when you find something it can scan it and take a picture for you as an example, auto-sort the things you've scanned, ect.It can scan all humans automatically and track them for you, displaying the hostile ones as red / purple.Of course none of the fun interesting stuff will be in it and it'll never be this advanced even though security cameras can already do this, YOU as a person can't do this. Only zog companies can.
>>537707127Speak whatever third world language you can. Just no more english. You look like an idiot to the first world
>>537707317You are unable to form sentences in English
Every day I work through AI.I'm getting a lot more done.I don't see how it's a scam, besides that currently it's probably cheaper than it costs.
>>537707359basically it works only in programmingother types of work will just get obstructed by usage of AIeven language translation (english, spanish, whatever you like) becomes unpredictable after AI jumps in
>>537702189What is a "sense" other than a constant flow of information input? That's what the cameras and microphones are there for.Artificial Intelligence doesn't lack senses, it lacks intelligence. It can mimic the output of humans, but it will never mimic the process used to reach that output.
