The AI LIE
HTML-код
- Опубликовано: 29 сен 2024
- AI is NOT taking over or getting smarterer... .
►► Digital Downloads ➜ christitus.com...
►► Reddit ➜ / christitustech
►► Chris Titus Tech ➜ / @christitustech
►► Twitch ➜ / christitustech
►► RUclips Early Access ➜ / @titustechtalk
By trade I am an autocorrecter. Is my job in danger?
You’re screwed my guy.
Oh you are the ones that will get leveled up because you already know how to use autocorrect and aren't scare of it. haha.
😂😂
*Autocorrecter
we're like 2 years into the "AI will take your job in 3 months" phase.
No, you're 2 years into the "AI will take your jobs in 5 years" phase. And it looks like we're well on track to see it happen.
@@compteprivefr did you watch the video at all or are you just huffing that AI copium?
Really, we're 60 years into the "AI will take your job in 20 years" phase. There's a long and rich history to the vast overestimating of machines.
@@roccociccone597 I watched it, and disagree.
Machine designed to do a vary pacific task, will always be made, the buzz word is AI. Will it take all our jobs, YES that's why its being designed. The excuse is "create more jobs" that not how industry works, it fundamentally must reduce the amount of people it costs to do a job, other wise they will just hire the human. No business will fire one guy use AI, and pay double for two people to maintain "the Machine", otherwise they will just hire the one person.
You are SO right! The exact same query on a google search will 99% of the time result in the same information spit out of a LLM.
I wouldn’t say google tho, it is long lost. Every other search engine has gotten a whole lot better. Might even be better at times.
Top comment if it's sarcasm
Well, you're right, but most artists, and most people in general, are not exceptional. Even if AI does average work at best, that's enough for most companies. The knock-on effect is that it's a problem for those exceptional individuals as well.
AI is essentially a search engine that has been indexed in many more areas, but with a specific cutoff date. It is very good at copy pasting other people's work. But because it doesn't actually learn, all it can do is imitate based on its indexed data. From all its "training" data, it gives you responses that best align with your request. In the packend, it converts your requests to computer commands.
YES! And for that is is useful, can you imagine not having going through book or Google pages to find information when you can infer immediately from the LLM that can somehow interpret questions.
This comment wouldn't meet training data standards. You're reducing something you don't understand down to your level for no other reason than to criticize with limited knowledge. You should learn how the technology works and what it's used for. It can copy us but it also does what we can't.
you will be amazed of the new reasoning models then
Been using Sonnet 3.5 and 4o a lot. It's good, but merely a tool.
@@TitusTechTalk Those are more aligned with "autocorrect on steroids" as linus called it. The o1-preview and the upcoming Orion model ~December are reasoners and will enable agentic workflows
@@TitusTechTalk You're a tool! 🤣 Just joking, I saw the opening and took it. But seriously, I think you are under selling the capabilities and maybe misunderstanding the technology a bit. Also, I think that someone that works in the area of admin and configuration, you want to be especially cognizant of these new technology's capabilities. Admin and configurations seems to be some of the easiest things to automate with these technologies. These areas are low on creativity and ingenuity. Unless you are developing new platforms or infrastructure technologies, which does require creativity, "AI" is coming for ya. A switch flipper, button pusher, and dial spinner seems to be an easy thing to replace. I expect admin and devops people to be the first people to go, in a a major way. IMO
I usually agree with you, and i used to think like this, but honestly, I've seen how o1 (strawberry) works with my code, not only improving but actually trying to optimize it and "thinking" of multiple ways of achieving the same thing in different ways. We are getting to the point where you can use AI to "think" for you and look for a different perspective. Hell, o1 was literally testing my bash scripts while "thinking". The future looks promising, i really believe we will get to a point where any user will be able to just type what they want, and the agent will build the code. I just saw the OmniGen project (OmniGen: Unified Image Generation) and I'm blown away by this. IDK I'm just very excited by all this cool tech.
I wonder how well o1 can scale its performance to large projects. It's one thing to write a Python script, but actually writing something like an OS from scratch seems infeasible with current LLM architecture. Maybe your experience indicates otherwise
@@RoboKestrel"your car can drive you to work, but not to space, maybe you should reconsider how useful your car is"
Correct. This was another boomer take. Even when offline these models are greater than the sum of their parts. People aren't spending billions of dollars on machinery for creating "better auto-correct." Five years ago you could relegate it to the realm of another search engine or algorithm, but that isn't true today.
AI is definitely coming a long way. At the end of the day, it's still just a tool. We also have to keep in mind we still should know more than the AI as it can still be wrong. It was for a N64 project and it didn't understand its limitations and tried to unroll loops especially.
It's good, don't get me wrong and helps having an extra review. But it's also important to know when the AI is wrong and not trust it 100%.
Being a soulless glorified autocorrect doesn't exclude it from becoming more useful than an average but qualified human to accomplish specific tasks. It's gonna happen.
You were totally going places, until you started talking about souls in artwork and music. It's as if words fail people as to why art and music cobbled together, don't move people to wonder and awe. It's because they lack cohesion. There is no unifying theme to "AI art". Be it in the composition, brushstroke or note clusters. Saying it'll never happen however, is a huge forecast that's bound to fail. A "hold my beer" waiting to happen.
Soul? Are we living in a fairy tail?
Not A.I its ACS by Titus 👍. Give A.I a soul and you Get Terminator : Zero 😂🤣🤣🤣.... Seriously it's not a Joke.
i am an ai screptic too, but i am kind of worried about it. the internet will be flooded with a lot of ai (low quality) content by using a lot of energy. will the internet get even worse?
It could be argued that all existing knowledge and learning is build on the shoulders of all the others that came before us in the natural course of history.
My 11 year old learns from me while she grows up... I'm hardly going to sue her when she's an adult for stealing all my knowledge, lol
LLM may not be A.I. But outside that what do you think about the rest of the field? No human is going to be better than AlphaGo Zero nor compute protein folding at a comparable rate. Humans can't function without language and are specialist in that regard. but at other tasks they are not so hot. Prediction of the properties of various compounds and materials might be where the main revolution occurs. And robots.
We've had protein folding algorithms for well over a decade. It's not revolutionary and hardly the result of the current wave of AI.
You look like Jerry from Rick and Morty.
AI is not auto-correct
AI is Auto Incorrect. Basically I see it as a TLDR; summary of everything and it's sometimes slightly incorrect, and often massively wrong.
never say never
I agree…but i hen there is a dollar to be made…people will tell you they can make a rock sentient …but we are not there yet …graphite is something that came with huge promises and delivered nothing.
Thank you for your insight and pacifying remarks on the current situation with AI and its possible prospects. However, your words revived in memory a couple of things (let my philosopher alter ego speak):
1) an article I read once in one of the highly popular American scientific magazines a few years ago, which claimed that AI will NOT be able to compromise our privacy, nor to steal our jobs. This claim is failing gradually and miserably, leaving behind tons of "conspiracy theorists", a diminishing number of "progressivists" as opposed to a growing number of desperate yet faithful people, especially in what they call the Global South;
2) the historical development of human civilisation and science for the "good of ALL people!", which have always been fast enough to invent a knife (nuke, rifle, ... whatever), without thinking first about the consequences for the rest of the humanity and bringing that "good" only for the selected few (the Pareto principle rulez!);
3) were the concept of progress true as the ultimate goodie for the whole humankind, why care about the cancer?
4) we are different from AI first and foremost because of our merciful side and love, including for "non-exceptional" artists and other creative people. Imagine what a completely pragmatic world based on practical benefits only would look like!..
People can do awesome things. But most people don´t do awesome things most of the time. Most stuff you find is mindless. Elevator music. Background art, designed to distract.
And I agree with you that AIs probably never will be able to do awesome things.
Still: We want to know: How can we do awesome things more often? (We have asked the similar question: "What - exactly - is art?" for ages)
Computers have been pushing the boundary of "only humans can do that" more and more. Chess, Go, now Text "creation", image "creation"...
So I see LLMs as a chance to learn more about ourselves: Where is that boundary - exactly? What is it - exactly - that only humans can do? We might even (indirectly) learn what it means to be a human - by "substracting" what AI can do from what we can do.
There is absolutely nothing humans can do that AI won't eventually be able to also do.
Sometimes I'll get stuck when I'm programming and need some inspiration, so I'll search around to see how others have solved the problem. Then I'll write my own code based on these ideas. I think AI is something like this, but with the power of computers to search vast amounts of data quickly. AI is very fast, but on the downside it doesn't understand anything being asked of it, rather it just pumps out the best pattern matching the question. For me, pattern matching on steroids would be another way of seeing it.
All this depends on how you butcher the definitions. Background would be important if you were going to be reasonably exact. I would also like to hear about LLM and translation. And then there is what relation does LLM have (if any) with those programs that show learning behavior to solve problems in science and engineering. A room temperature superconductor or a cheap higher efficiency photovoltaic cell would be worth trillions of dollars.
The real "AI" will be based off quantum computing, where it won't be trained on vast amounts of copyrighted data but on itself. Watch Michio Kaku Quantum Computing
The new o1 is kinda cool tho, so when asked something, on the background they generate thousands of tokens trying to simulate "human thinking", before giving their response. Personally, aside of the autocorrect term for current LLM, I always thought it would be impossible to synthesize human intelligence, but this new paradigm do open the door for new possibilities, if prior LLM can copy paste human work, what if they can do the same to how humans think. o1 stills struggle with things not on their dataset, and some thing GPT-4 gets right o1 gets wrong, I guess because if before they were confident they were right, now with this new chain of thought baked in that make them arrive to whatever answer it is, it also acts as their own "proof" they're "right", even when not.
My problem with using AI to shortcut my way to building / coding / whatever is that I will literally *never* learn as a result of it.
So I just can't do it. I'm the same with all this f'kin GBs and GBs in size 'generators' and stuff that a lot of web technologies tell you to use in their docs ... I don't want to yeoman ... I want to know how to build the damnable hello bloody world!!
Thanks for the video. A good book about todays AI and how it basically works is "Smart until it's dumb" written by Emmanuel Maggiori
So true about creativity and art. When I see and hear these imagees and music now, it just feels repetitive - the same thing I've seen a hundred times already. The first time, it seemed impressive. But they all look the same derivative blah.
Automated thinking and reasoning. Much better than non-existent organic thinking and reasoning.🎉
I'm afraid you're overestimating people when it comes to how little they are fine with when it comes to AI outputs,
in the long run people becoming more and more dumb and originality being shamed/undevalued,
AI hype is obviously makreting/sales stuff, but again, people will jump onto that nonsense,
and once AI and automation reaches certain level, human input/output will become blacklisted by design in worst case scenario
I want LLMs to ingest application documentation and then be an easy way to pull out relevant parts in register language.
And then if I'm not getting what I want I want the llm to be able to spit out the exact phrase and location of the response information
Is AI really just autocorrect? If it were that simple, why didn't it contribute to its creation or development, rather than being a glorified Windows app cleaner?
By the way, I asked Llama 3.1 on Ollama to give its response about what you said.
Large Language Models, are simply a large if else statement with billions of entries. So its more of a glorified large database.
I am not sure if I totally agree with this. Just saying that this is autocorrect on steroids is huge understatement.
I agree that AI is currently hugely over hyped. But it is a lot more than what you are saying.
AI doesnt have emotion it can fake it kind of but with out emotion it would be hard to fool someone maybe written messages can possibly fool people but without emotion there is no true understanding
I personally hate those videos with ChatGPT scripts and ai voice overs they are terrible soul less dribble..
This is some nuclear level copium.
Synthetic data is a thing my dude.
AI is a thing a while now but i still didn't find use for me. It's useless for me.
Can't wait for the apology video three months later
I'll have "AI" make it.
@@TitusTechTalk it'd be so soulless lol
auto-correct on steroids - came to the same conclusion. nice
They think they can program people 😅
Garbage In / Garbage Out
we all gona die bro, it took our jawbs
💯 🎉👍👏🙌
It's sad that tech bro's turned AI into nothing more than a buzzword.
LLM's have never and will never be AI.
When our robot overlords assume command, I pray they go after the Tech Bros first.
I'm not religious.
What about Tech Support jobs?
💯 agreed. AI is a buzzword Stupid people use to try to sound smart.
yeh this
So much cope.
AI is a misrepresented term used with confidence just like how developers use the term "refactor". 😂
F* it! We’re doing it live!
Auto correct on steroids that uses and insane amount of power, which is driving up emissions which is pushing us closer to extinction. It's a shame that its doing all this damage for mere enhanced auto correct.
Weird how everything the US govt wants to control is gonna blow up the world and kill everyone. AI, bitcoin, the Iraqi govt, agriculture, all types of scalable power plants.
It's just really unfortunate, thank god the US govt is here to protect us from them.
"AI" will never create Dadaism
...not right.
pillaged.. like Vikings?
W take
And machine learning is glorified statistics.
Ai What a normal google search use to be before they switched to plugging adverts?
Google from pre-2012 was a lot closer to AI than what they pass of as a search engine today.
There are two things at play here. First one is the AI look and feel. Even though the generated media are distinct, they have a uniform feel to them. They don’t have the progression/storytelling sapient arts tend to have. I think this could be possibly overcome in 10-30 years.
The second part is more intrinsic to the process of consuming art. I don’t think we can appreciate art knowing it has no intention behind it. I don’t think this can be overcome until the generative models are more believably sentient. I don’t think this can happen within this century.
תודה!
I like to use AI art mainly for posing for my own drawling, not tracing as i don't. Just reference like Pinterest.
Learn anatomy.
@@araa5184
I have learned it in school. though I'll tell ya art class sucked, my teacher was an old grunt who didn't believe in my cognative issues so I wasn't given any assaistance and I struggled hard. I found that this helps since I am a visual learner and I do better when I can see what I am doing. so poses with AI is similar to those stick guys artis use in art class. you can even get them at the dollar store. I have one but I doesnt help as much as having it digital.
now I am not tracing, that's no fun and makes me feel like it's not my art, it's plagiarism. I'm basically telling it
"a masculan model without any detail, siting down on the edge of a roof of a house, smoking a citgarret, with power lines and sunset in the background"
as an example. I got the pose and the foundation I can use to make my own. it's never 1:1 with what the AI has made. becuase that's no fun
@@Yotes_ I see, great that it works for you
@@araa5184
😁
Tldr lol no timestamps
LLM being dumb and is autocorrect. He basically summed it up first 30 seconds.
ai is the worst it will ever be right now
AI companies can weasel around until some judge in the US decides that collecting data and storing it in the form of AI model's weights with the purpose of creating and distributing derivative works is no different from collecting data and storing it in other digital forms for the same purposes. There were published cases when AI models were trained so well they reproduced copyrighted texts with all the original's mistakes and even copyright notices, images with author's signatures etc. Once some good lawyers properly bring these cases to the court, the story of the modern day "robber barons" will be over. Unless some politicians (may they live for a hundred years and rot 1 millimeter per day) decide that now some companies are "too big to fail".
AI is summed up as:
Ordinary person: _Create for me an adventure with this type of rad orc villain!_
AI: _I'm sorry, but I'm not permitted to assist with that._
autocorrect on steroids is spot on
AI, ML, LLM the name doesn't matter. Whether or not it's truely philosophically creative or original doesn't matter. This emerging technology, which still is in it's absolute infancy will break the world. "Some jobs will be lost", yeah, right... Lets reconvene in a few years when all thats left are robots and influencers 😀
I don't think this will age well; much of what you said is subjective. If I wrote a great song that would be a top 10 hit back in the 90's, but I told you it was Ai, you would think it lost "soul". And they still haven't found the end, they have a long way to go before it can be considered "mature" software. The art only looks fake for now. Notice how it looks much better than it did last year. Eventually it will look perfect.
I can follow you on this "AI is badly termed, it's just autocorrect on steroids" or properly using the term autoregressive prediction or whatever. But when you get into this whole diatribe about soul and art... my eyes glaze over. It's honestly already good enough to produce things if people didn't know about AI at all, they'd have no clue an AI did it, and they wouldn't be saying any of that crap about it
There is so much art that is produced where people didn't spend much real time or thought on it, but are just doing it for the paycheck. And there is nothing wrong with that. There are tons of mediocre songs made by great bands. They don't get many track purchases. AI tends to play to the averages in code or elsewhere. Prompt engineering helps reduce the scope of what it selects and thus improves the particular set of things it's averaging from in a sense.
Don't get me wrong, I think the whole "art has soul" thing has some merit, but I also think AI copies the "soul" in some instances pretty well. If you haven't seen that, it's because you don't want to see it. None of this means I think humans are going to be completely replaced doing this. But I also think a lot of humans doing art are going to be pretending to do it while letting the AI do its thing. And all the smart people think they'll know all of it when they see it, but they won't.
Humans see artistry and awe in nature itself, in the stars, in the forming of clouds. These things have "soul"? I doubt it. We largely see the patterns we want to see. We are in search of meaningful things, whether they are there or not sometimes.
I think AI could actually be useful if people are selective while feeding it. But this "let's gobble up the whole net for the maximum sample size possible." approach has some laughable results sometimes.
You nailed it
Sounds like AI wrote this script.
There is no script.
This is either delusional or wishing thinking. Most of this is nonsense
We are not allowed to steal other people's art but they can
"AI art is soulless and bland" yes, when its being controlled by people that have no artistic skills, taste or vision. keep coping i guess
not taking art criticism from the "my specialty is youtube videos" guy
I like your vids but some of this is straight-up boomer-tier. Your complaints about AI music are VERY reminiscent of what I heard about Hip Hop in the 80s. I do agree that the term AI shouldn't be used but an LLM and the generative AI we have is far beyond an autocorrect. It's generative which is the massive leap forward that we just took.
You get it dude. He has a point with the AI art but he couldn't be more wrong about it just being autocorrect although that is a part of how it works. AI is genuinely creative. Ilya has spoken about this before.
It genuinely seem like he doesn't know how even AI chess works,in the very beginning they just tried to feed it all the best move but kept losing because any human can just use subpar or unexpected move to trick it, spoiler they built two different models so it auto self correct it self, like a human brain and yes to be clear we all agree its not human just slightly modded off the idea.
Auto correct ... LOVE IT!
Wow you are so wrong about this topic!!!!
Dude's gotta have some copium. People just aren't ready yet for the reality of AI.
on the spot titus
I totally agree but i think cursor is a good depiction of the direct of possibilities. None of the cursor knockoff work as well, they aren't even comparable in quality. That tool is absolutely a "Tool" and not a replacement. I don't think its going to stop at LLM's though. True AI will come in the future. There will be a day when AI "Tools" become replacements.
LLMs are way more than autocorrect. Local LLMs do a better job than Alphabets search engine. If anything LLMs are advanced chatbots with built in search engines!
Thank you!!!
Thats exactly what I've been telling my friends for years now. LLMs are NOT AI.
There's a subset of AI bros that have come up with the excuse that "nothing is original anyway, people don't create original ideas, they only steal and recycle ideas and things from their environments and each other" to justify AI from a legal as well as philosophical and moral perspective.
And yes, they *actually* believe this.
All it takes to dismantle this argument is to point at things like Mayan and ancient Japanese architectures and arts. Who stole the ideas behind those from what or who? There are so many more examples than those two, sure, but those are two particularly standout examples.
I agree 100%. Thank you for expressing MY thoughts on youtube. Wha
This is not entirely correct. Autocomplete tools can count how many Rs are in Strawberry.
You don't understand LLMs. It is not auto-correct on steroids.
AI. Another Indian?
I agree, nothing intellegence about AI. Its just fancy search database.
I completely agree
That's right
okay, you lost me at "tattoo artist"
I guess I should have said she also was featured in gallery for a bunch of art as well. Has a very impressive resume for an artist and sold around 40-50 art pieces before getting in to tattoo and people will travel out of state to see her. So I consider that an artist.
Maybe not a digital artist with an ipad and photoshop, but are they an artist? Hard to say.
When people say 9/10s of the law is ownership, they mean 90% of laws written are about who owns what and what they and other people are allowed to do with it. These AI take pieces of other peoples work and stitch it together into a plagiarized word salad or mash of art. Sooner or later they will be sued out of existence like you said.