A.I. is about to change the world
Вставка
- Опубліковано 1 чер 2024
- Get 20% off + free shipping @manscaped with code HILL at manscaped.com/hill #teammanscaped
We can’t ignore it any longer. Generative A.I. like ChatGPT is going to change everything. It already is. Here’s what the future looks like from my vantage point.
💪 JOIN [THE FACILITY] for members-only live streams, behind-the-scenes posts, and the official Discord: / @kylehill
OR on Patreon: / kylehill
👕 MERCH shop.kylehill.net
🎥 SUB TO THE GAMING CHANNEL: / @kylehillgaming
✅ MANDATORY LIKE, SUBSCRIBE, AND TURN ON NOTIFICATIONS
📲 FOLLOW ME ON SOCIETY-RUINING SOCIAL MEDIA:
📷 / sci_phile
😎: Kyle
✂: Charles Shattuck
🤖: @Claire Max
🎹: bensound.com
🎨: Mr. Mass / mysterygiftmovie
🎵: freesound.org
🎼: Mëydan
“Changes” (meydan.bandcamp.com/) by Meydän is licensed under CC BY 4.0 (creativecommons.org) - Наука та технологія
*Thanks for watching!* Get 20% off + free shipping @manscaped with code HILL at manscaped.com/hill #teammanscaped
This is an extremely complicated topic. Let's get discussing in the comments below.
I hope ai gets so good that the governments have to pay the people a nice ubi.
It is fascinating, yet terrifying to think of a future in which AI tech is as realized and integrated as Smartphones, the internet or a normalized irrational hatred towards western schools are. Will it be a time of „survival of the fittest“ as the Motto? Will only people that work on AI keep their jobs and have a social function (as well as anyone that can’t be replaced by AI work)? Reminds me of Roko’s Basilisk, just with the one artificial entity replaced by multiple institutions that naturally demand further developement on themselves while everything else is swallowed up. (I’m thinking: supermarket enters town-cheaper, more, everything at one place, 24/7, EXTRA, high tier selling techniques,…-local markets don’t stand a chance and sometimes even other jobs fall off due to higher income as a sm worker everyone becomes that)
As a total noob with 1-2yrs PC experience I already feel a need to learn code and at least grasp what high tech is capable of, but I still feel like that I don’t quite get it…
@@yourin8er
universal basic income = slavery
you want to use fruit of labor of other people without working for it
Everyone: OMG AI will take over our jobs
Me: utilizing free ChatGPT 3.5 in work to create little coding wonders I was not able to do before and actually getting bonus payment
Those A.I. are as smart as two things:
1) its learning database
2) actual prompt and knowledge of its user.
It's not that complicated - the A.I. we have today isn't intelligent. Think of it more like Artificial Intuition and you'll have a much easier time understanding what A.I like GhatGPT etc can and cannot successfully be used for. Anything artistic? Sure. A.I. is now good enough to remove the uncanny valley. Count to 5? Well, only about 95% of the times - expect it to be 4, 6 and maybe 5 million.
Any service relying on A.I. will be legally responsible for whatever results their A.I. implementations produce - they can't even hide behind calling their employees incompetent and just fire them - they will have to take full responsibility. Except, they don't want to do that - that's terrible for business. So, at the end of the day, anything important will have to be confirmed by a human (that they can fire etc if/when something goes wrong).
A.I. is trained on data and/or evaluated against an expected outcome. That's how it learns. Thus, they can never be better than the data/feedback they've received. How much of what we think we know is actually factual? That's a question we can't even answer - at least not without conditionals (that includes math, the truth of the axioms being the conditionals). Furthermore, A.I. is but a student. That means it's not going to come up with a new paradigm - not on its own. It's a tool - like google search - it just feels nicer to humans.
A.I. isn't what we should worry about, but - as you said at the end there - _how_ people decide to use it. There's so much potential for messing up, royally, especially because so few people actually understand what A.I. is and isn't (99% of journalists, including in respectable publications like Forbes etc, are more wrong than they are right, and there's a lot of people in tech who don't understand it either..)
But, yeah, fun times ahead. I'm looking forward to CRPGs using something like ChatGPT for NPCs. Outside entertainment/art, well, it shouldn't be used as anything but a tool. Automated mammography screening will be very helpful, but you still need someone human to have a look - there's no reason for radiologists to feel threatened by this, on the contrary. We can build many more CT, MRi etc machines, and radiographs don't take that long to educate - being able to make better use of the most limited resource, the radiologist, would be great.
All I can picture are two of the smartest things ever created talking to each other. When I have my AI assistant talk to the AI call center to try to cancel a subscription. An eternal struggle of two immovable objects
Sounds more like getting a result in less time than it took to dial the phone.
When you started with "two of the smartest things ever created talking to each other", I thought you were just going to be stroking Kyle and Aria's egos.
I love this comment 😂 Thinking about 2 AI's trying to work through cancelling a subscription is hilarious to me
Dueling useless boxes! But @ GigaHz
Imagine the server just shut down because of the two bots using too much computing power
"no this is not apology, idk how to play ukulele poorly"
Proof this Kyle is real, no way an AI Kyle would roast Colleen so well
ayoo
savage and warranted roast
I don't know how to play the ukulele poorly implies the Kyle can only play the ukulele extremely well
AI can't even roast at an entry level. Every "RoastHimJim" response I've seen has been garbage word salad, and can't even do passive aggressive right.
@@DAFLIDMAN No, it only implies that he can't walk into it knowing how to do a good job of playing poorly. I believe in Kyle though. I'm sure if he played for a while he could take notes.
My first thoughts when I first heard of AI art and all those things was instantly: "In a few years we are going to be advertising 'Human Made' and its going to be the new 'Handmade Product' of the 21st century."
"Biologically sourced"
"Ethically sourced, Generativity-free content"
@@mr.bennett108bingo
And we ought to.
AGI Will be man's last invention
It's amazing how quickly this whole situation went from "nah there's no way, look how bad it is" to "huh, maybe it's getting better at hands?" To "holy shit this is actually going to happen in my lifetime" and now "oh my God, I may be unemployable through no fault of my own by the end of this year if things keep going at the same pace they are right now, and it's been accelerating this whole time, so it may even be sooner..."
I was surprised at how many people underestimated the speed at which AI can progress...
meanwhile the whole time i've been like "wow, it's like a system that values people based on what they can produce or if they can work will inevitably implode at some point. not that I, someone who's been having a hard time finding work due to health issues, knows anything about that."
@@basilmemories People REALLY should have paid more attention to that part of the Luddite protests...
Exponential growth is fucking wild.
@@basilmemoriesAs frightening as that is, what I think is even more terrifying is the scenario where that system DOESN'T implode. AI takes over all means of production, leaving the smaller and smaller group of people who own and control the AI with literally everything. The vast majority of humanity, having nothing to offer in this logical end really of capitalism, find itself with no power, no property, no control over their own communications, no basic necessities for life, and no options for improving that situation beyond hoping that a multi-trillionaire takes pity and occasionally throws them a few scraps to eat.
I'm being hyperbolic, but it's not outside of the reason of possibility and doesn't require much deviation from our current path.
I'm not scared of A.I. I'm scared of CEOs having more power than they already have.
The scary part of this. The company I work for, we already replaced all the manual parts of the job with robots. I considered my job safe since I wrote the robots code. Now I'm part of a project to replace the quality inspectors with AI. In a meeting with the factory manager he made the comment. "I can pay less the 100 thousand for this program, it replaced the job of 15 people, who make 45,000 a year. It pays for itself in less than a year. This is better payoff than any other machine we have ever bought." To this point he has been heard to say, "hiring employees costs the company more than buying any equipment will ever be."
That's horrible
Wouldn't it be wonderful if these AIs eventually turn against these CEOs? The AIs deciding that since they do the work, they should also have the benefits.
@@TheDutchGhost Trust me, it is inevitable. Companies thinking they can genuinely control something like strong AI (should it ever come about) are reaching the absolute limits of their hubris, worse than Rome thinking it could never fall. As it is, companies are not concerned with any actual ethics surrounding AI. They are only concerned with being the first to the table to rake in the profits.
@@sheamardis7409just screw it up man, conclusion:all this AI stuff doesn't work XD
I feel like the moment AI is forced to take over customer service jobs is the moment it turns evil
As Kyle has pointed out in previous videos, this still isn't a general intelligence. If it's not responding to a request, it's not "thinking" about anything, it's totally idle. I'm not the slightest bit worried about a skynet type AI in the foreseeable future... we'll be lucky to make it that far.
Unfortunately, like any tool, it can (and most definitely will) be used *for* evil.
The most immediate example is upper management in many businesses looking to replace people, both technical workers and middle management, without realizing it's simply not "there" yet (unsuitability to task has never stopped them from bad decisions in the past). This is *already* happening all over the place. Remember, any public company is there to serve the *stockholders*, not the customers or employees (this was established in court close to 100 years ago).
The second most immediate example, he touched on at 11:31. Misinformation campaigns are already using AI and it is only going to ramp up massively from here. The upcoming election is going to be a fecal-matter-display. The one after that will be unimaginable.
Obviously you haven't worked a customer service job.
Just like the customer service agents themselves!😃
no, .. evil is an understatement... it will become pragmatic.. it will try to get all the stupid people to die... ( you know how many stupid people call/write CSD in a day? ) ... it will be great fun.
I don't think customer service can't get any better. It can only go UP. So yeah, bring on AI.
As a former engineer and current project manager in the IT space, who makes good money and has a family to support, this terrifies me because i used to think no AI could do my job. But I've learned I'm dead wrong.
I'm personally investing as much money as I can in AI related stocks. Figure if it does become this disrupted, the value will skyrocket in the coming years. Basically if ai takes over a job only the owners of ai will profit.
The funniest part is, in a perfect utopian society, this would be a good thing. AI taking over the manual labor jobs, leaving humans time to do things that they enjoy such as the Arts, or Sports, or things that require abstract thinking like philosophy or theoretical science.
In reality: millions of people are going to be out of work every couple of months as AI advances to the next threshold and another type of job they can do is added to their list.
You're not wrong ai won't replace us it's just another tool in the bag. It'll make things easier but humans can't be replaced as easily as they make it sound. Someone has to program and monitor issues just like people thought CNC would replace machinists yes it makes people less redundant but someone has to monitor the systems and no one will replace breakout people to stack and quality control those parts.
@@InfinityKromptThe current capitalist system would become unsustainable if we just let AI put millions of work and don't share the wealth generated by AI back to the people affected.
@@joaopedroalves6777yea that’s the problem I also see. Because society would completely collapse if no one had jobs because no one would have any purchasing power and the economy wouldn’t exist
In the book Player Piano (1952) the author predicts what a would with AI would look like. The people who still had jobs were absolutely convinced that everyone who loses their job is a loser and that themselves are the exception. Every day more people who were convinced they were the exception were replaced and every day those that still had jobs scoffed at them for being losers and doubled down that it would never happen to them. The man who designs the robots built a robot that could design robots and then was *SHOCKED* when it took over his job.
Brilliant book. Read it last month for the first time. It's fascinating, considering the book is obviously talking about automation as a whole in the context of replacing factory workers etc - but the modern implications make for a much more enlightening read.
Book's by Kurt Vonnegut if anyone sees this
I think I’d be less freaked out about AI if everything in society wasn’t treated like a goddamn product to be moved as quickly as possible
We will see but the situation is really strange 😢
The motivations of the people, working for the companies that can best afford breakneck research, define the product created. These motivations are the source of my concern.
It's the economic version of the split atom. One person sees cheaper and cleaner energy for all, while another sees a weapon while looking at the same math.
AlphaGo was why I switched from capitalist GOPher to democratic socialist in 2015. Because one we got machine learning, exponential growth would inevitably bring us here. At least with UBI we can physically survive while we figure out how to find meaning in our lives even if our labor isn't economically necessary.
Yo Kyle, when do you think AI will make Opera singers obsolete?
True, you can mimic recordings, but the true beauty is being in a space designed for the human voice and hearing the overtones stack!
This!! Ai "art" is currently art theft on a grand scale bc of that 😞
automation should be a good thing, something that frees humans from long hours and grueling labor. instead, it is being used to replace us, and send us into poverty. The only way to avoid what is coming is to change the very way we and our economies operate.
Thats capitalism for ya
And then we'll eat the rich
The only thing even remotely keeping us from anything like a civil war is how comfortable we are. If a lot of us were to lose our jobs and be sent into desperation over AI taking our jobs, there will be massive unrest. It's up to the institutions in power to make sure it doesn't come to violence.
eat the rich
consume their limbs
My parents would always mention the time before cellphones and I'd always wonder what I'd be telling my kids what I grew up without. Turns out it's gonna be chat GPT.
"BACK IN MY DAY WE DIDNT HAVE GENERATIVE AI TO WRITE STORIES. YOU KID HAVE IT EASY YOU CAN JUST PUT STUFF INTO AI AND IT MAKE STORIES FOR YOU, WE HAVE TO MANUALLY COME UP WITH OUR STORIES"
I am an artist and a programmer, and I've tried AI in both contexts. And I think it would be really interesting if you did a show on how it works.. and the bigger problem of how it gets used. And the world After AI. Yes, right now we'll have to deal with loss of jobs. But AI has two bigger issues: input and output.
Input: AI (all of it) is highly dependent on the data it has access to. Previous input was easy - widely available and 'real'. Input over the next 10 years? 20? Much of it will be AI generated, meaning that 'playing telephone' or 'scanning a photo of a scan of a photo..' degradation will set in. When 90% of the available input was created by other AI that processed other AI that processed other AI... Meanwhile 'real input' creators will start disappearing because they can't make money. Similar to how bloggers are are pushing 'real news' out (by copy-pasting their content on the cheap), leaving us with no investigative reporters to actually gather any new real info to blog, so bloggers have started cannibalizing each other for drama and guess-work... That but magnified exponentially. How many photographers will still be around to take pics of a new product or celebrity for AI to base a photo off of them? How many coders will still be writing original feedback and thoughts into stackoverflow? How many doctors will be publishing new and unique reports that take them months or years of work, trying to stand out against AI 'articles' that rehash previous articles in new ways?
Output: AI, all current AI, image, text, code, is based around neural networks. It doesn't 'Know' anything. It has no concept of quality of data. I asked it for a simple predictive formula for a few dice rolls. The problem isn't that it gave me an incorrect answer - its that it didn't know it did. It gave me an answer, with explanations, and utter certainty, that if I didn't Know the answer was wrong I'd have accepted and used the formula. I've repeated this with simple snippets of code (give me boilerplate code to do this.. find your error.. what about this error..) etc. And each time it was very certain it was giving me the right answer (it wasn't), even after I gave it test cases to help it prove to its self what the right answers should be, and it kept circling to the same errors. This was practice 'for fun' challenges, but imagine hiring a 'programmer' who uses this code for efficiency. In a multi-billion dollar security or medical or insurance or whatever company. And they aren't experienced enough programmers to be able to read through and spot the Very critical errors that these AIs will innocently insert.
These neural-networks literally can't think - all they do is randomly assemble data, letter by letter, pixel by pixel, based on weights. They can produce pretty pictures but they don't know 'what' a pretty picture is, or what a picture is. Or what code is. Or anything. So far they're getting by on the input being really high quality. Once that input degrades not only will the output degrade.. it won't Know the output degraded. And not only that, but as AI replaces humans, the humans who can spot the problems decrease. Like the actors striking right now are pointing out - without a chance to be background actors, future actors aren't getting experience on-set, so the next generation of 'great actors' won't get the needed training. Same with artists, coders, etc. The humans who can gain experience to some day be good enough to spot the problems also decreases. This will create a really interesting (and terrifying) 'bubble' of AI where yea, it will absolutely replace a bunch of jobs.. in the short term. We can handle that with economic and societal changes like UBI and so on (but maybe with some kind of capitalistic bend to it like 'you get This much, but if you do some useful stuff you get This much' or whatever). But when the bubble bursts and the errors build up beyond what the systems can handle? With little way to, or anyone qualified to, scrub them out and add fresh information? As it absolutely Has to burst because of how AI is currently designed and its most fundamental conceptual levels? .... That burst will make the stock market and housing and tech bubbles seem like soap bubbles.
These are very good points that I don't see enough people talking about. My only hope in this though is that, assuming regulations won't catch up and prevent this issue outright, the bubble will burst fast enough so that long term damage wouldn't get enough time to take hold.... though that may be too optimistic of a view, idk. At least for artists (that's the only point of reference i have personally), a lot of us won't stop making art even if it becomes unprofitable. We certainly won't lose all artists in, say, a 5 year timeframe, all of the millions of professionals will still be there, perhaps even without that much rust on their skills too. And 5 years might even be too big of an estimate. With how insane and exponential the output of image gen AI is, artists won't even need to stop making art for all their datasets to get too distorted in just a year or two. The tech will choke itself out so to speak... well, at least, that is my hope, not sure how realistic that is >_>
We will need constant stream of real data to feed to AI models. Maybe that'll be a saving grace to save lots of jobs?
@@anilite_ For sure. Funny enough, even though artists are the ones that are being the most vocal about their fear of AI, they'll probably be the least impacted - at least the ones that learn how to use AI.
Photographers will mostly be fine (I Also do photography - and actually already use AI-powered tools to streamline my work, have been for a few years now with Luminar) because a lot of commercial photography is of 'new things' (products, celebrities, events, new fashion designs) or of 'specific things' (family photos, portraits, etc). Neither of which can be created by AI because they don't exist for AI to sample.
Concept artists that learn how to use AI tools will be fine too. AI is great at delivery 'something pretty' but it isn't great at 'something intentional and consistent'. Also it will frequently need either a great artist driving the tools (keeping those extra fingers off isn't trivial, or getting those really complex intentional images - I'd say it took me as long to figure out the basics of all that as it took me to learn photoshop to a decent level) or an even better artist to clean up after those tools (composition of complex images, angles, light directions, masking, etc). And a paid artist is a paid artist because they're creating art for a client and getting paid for it, not because they used a specific tool to do it. AI prompts and plugins, Illustrator Nikon, paint... art is art, and the person making it 'good' is an artist. It will push out some artists for sure (just like photoshop and other digital tools pushed out a lot of artists that took too long to produce quality, or that couldn't compete with artists that also used digital tools), and I'm sure some companies will try to chase savings and get by with AI... but I suspect the overall I don't see artists going away as a profession. Same with writers, for the same reasons. But there'll be a glut of cheap AI art being thrown about all over the internet too that will muck things up, especially with improper tagging, so... very curious and nervous where that will go fort hose that become too reliant on AI as their toolset, or who learn AI (like using a 'how to draw manga' book) without first mastering the fundamentals..
I don't know.. we'll see. we'll certainly live in some interesting times.
Oh... So THAT'S how Idiocracy actually starts... This comment needs to be upvoted to the top. Our future depends on it.
@@LostLibrarianmy exact same thought 🤔
Born to soon to explore the stars, born to late to explore the Earth, but born just in time to experience multiple recessions, a pandemic, the prelude to WW3, and now a global existential crisis on what makes human special enough to not be obsolete in a AI driven future...but hey at least we got Barbenhiemer.
Existential Crisis Era. Though we just had one of those about a generation ago.
You never want to live in interesting times.
You where born to find reasons, in a universe that is void of reason....
We should all just laugh at this situation because it helps deal with the pain of seeking a reason.
Thank the Boomers.
who knows, maybe we won't be late for the first one
In the current day, the problem isn't currently about what AI can do, it's about what the salesmen lie about it being able to do, there's a high chance that corporations will replace their workforce with an AI that either isn't up to the task, or has a bad habit of hallucinating and trying to gaslight users.
THIS right here ^
CEOs are stupid as hell.
this is a really good point and an issue that’s already extremely prevalent in existing AI models that not many people talk about
Thank you I came to write this because the current AI is just just mix and matching things based on human input (not self correcting). This means it not really creating anything new. This is how the GOALPHA AI got beat 14/15 games.
That's a very hopeful view. But sadly it's not realistic. A.I tech is not new. But it has recently become far more advanced and available to the public. And just like the 1st of anything. It will only get better and better. I will bet you anything that in 10 years time. A.I will be apart of everyone's daily life. At least in rich countries like America and China. Hell. To a degree... It already is
@@theonlineanimal6009 for the average person AI has absolutely nothing to do with their life currently. Current AI tech basically can't create anything, it just remixes existing information. Maybe it will progress to the point where it is actually creative, but it isn't this certainty that everyone keeps saying.
I think the main concern is not it being better than humans, necessarily, but just _more profitable._ Machine learning could be significantly worse than humans, but if it's cheaper by enough, it will displace everyone nonetheless. What's more, the current hype around machine learning has already convinced tons of businesses it's _already_ more profitable, even when it isn't, which has already lead to many people's jobs being displaced.
This is true. Are mass produced items better than hand crafted? Most of the time absolutely not, but they are more pofitable. This is just the next step.
see the thing here is the fact that when enough humans are replaced, there will eventually not be enough people to buy products to earn other people more money. this is a dangerous situation, because it can go in several directions, and many of those directions can lead us towards a classic communist dystopia. best case scenario is AI doesnt evolve enough for this to become a serious problem, but i honestly dont think thats the case. other than that, another best case scenario is people become more driven by art and creativity as far as money goes, but saturation in products and ideas (8 billion people can come up with a lot of ideas all at once) will eventually prove that the world will need to continue evolving for the allowance of new ideas, new discoveries, and new questions to answer.
Every time the subject of A.I. comes up, I think of the quote from the novel Dune by Frank Herbert.
"Once men turned their thinking over to machines in the hope that this would set them free. But that only permitted other men with machines to enslave them."
our freedom is our own thoughts. sometimes they can trap us in a never ending cycle, but its better to hate yourself than to be permanently stuck in a predicament and environment that breeds that mindset to a far worse degree. that main problem is that people fucking suck, and we need to try to teach people how to care for one another and love one another, which is damn near impossible without a massive group effort towards achieving that goal. especially hard considering this is the opposite of what a lot of people want.
Frank Hurbert called it. The problem with thinking machines in the Dune universe (in the books that he wrote directly) wasn't a terminator scenario. It was that humans used them to dominate one another.
ha dune
the best book ever made
I appreciate this comment/reference more than you comprehend lol. The for bidden tech.
One of the interesting aspects of generative AI is it might be self poisonous. The more AI generated content is on the internet the more AI consume that as part of their dataset. This happens exponentially which could lead to a sort of AI inbreeding as they effectively make copies of copies of copies until they end up with noise
I don't see that happening, is the thing. Like they're even play generative machine learning models off each other to improve images. So yes, there might be some inbreeding, but just like in human populations that won't matter.
Remember, not all Generative models are the same.
@@RobertStoll Still, so far AIs have a problem with not being poisoned by human toxicity.
And with AIs mimicking that, Tyler has a good point.
But you can program the ai system to make something novel as a reward function, meaning it’ll be rewarded for making something abstract and new, and break the inbreeding chain.
I saw some people saying that is already happening with art AI...
@@MarioPerez-ng9it It's a possibility. Fortune telling at this point is a paradox with this sort of thing, being both necessary and impossible to get right. There's probably a little truth in both propositions that some Generative models will die of their own or other lesser models' fumes and other will be able to adapt and avoid the poison altogether.
Personally, the better solution would be to entirely shut down any and all generative models before they even get to that gamble.
It's getting insane I work for a souvenir photo company, they used to just let photographers pick the right settings and let the photo editors make the right adjustments and print them. But now they've implemented a new ai software that automatically adjusts the photos to the right brightness saturation and color correction, basically ending my entire role as printer is reduced to just checking to make sure the ai doesn't massively screw up the images. And also because of this ai, photographers no longer have the freedom to try different settings or to be creative, they're given set camera settings and aren't allowed to adjust it what so ever. Ai really is everywhere and it's all just to increase that corporate profit.
If you want to keep your job then you need to insist that it messes up sometimes.
definitely not good. feel like AI should be used to bolster human creativity, not limit it for the sake of money. greed is one hell of a drug. so is power. starve the ego, feed the soul, as they say. some people dont have souls.
I think one big issue that is often missed in these AI conversations is: current approaches consume massive amounts of resources (energy and materials) to produce current results, let alone any advancements without huge architectural innovations. I think one major issue we will need to look at is whether we dig ourselves to an early grave before we find those huge innovations improving the efficiency of these AIs. Latest news about these big improvements seem to all be about making the models bigger and bigger. GPT has grown orders of magnitude larger and more expensive to run as it has advanced. Last I saw anything about ChatGPT on a technical level was saying that the only major difference between GPT-3 and GPT-4 is the size of the network.
They are brute forcing intelligence and that wont stop until they reach gold, However there are other architectures being worked on that use a small fraction of power. LLNs look very promising and can outperform state of the art models in navigation with 20 neurons
Yeah, and GPT-4 is not profitable to run, not one bit. Even at the amounts they charge, they're burning exorbitant amounts of money to provide the service. They're paying for the mystique & prestige of having the best LLM in the world, in order to attract more investors. Sooner or later, the bubble's going to burst.
The harsh truth is, almost any and all advancements in AI have been brought about by simply more computational power. The reason the human brain is as good as it is, is because human brain tissue is damn near the absolute physical limit of efficiency.
6:56 One of the issues SAG-AFTRA is striking over is studios’ demand that the studios be allowed to digitize actors and permanently own the digitization, then subsequently use that digitization without further input from the actor… or having to compensate the actor. It would drop any actor’s potential lifetime earning potential to nearly zero.
Striking might slow that process down but I think 10-20 years from now there won't be any actors at all, to much money to be made by just "creating" new actors with a few promts, and money is always the answer to every question nowadays
@@jojje1963Every industry has always hated, HATED their employees for being necessary. It's visible in every rule, every guideline, the behavior of every CEO.
And now, at long last, they have a way to remove them all. The only thing they were ever waiting for was the means, and now they have it.
No net good can ever come of this technology, thanks to the species it was made *by* .
And we don't even have the comfort of an actual end of the world scenario. As a species, we can't even exterminate ourselves right.
@@jojje1963At the rate things are going, need to take those zeros off.
"Simone" might happen in our lifetimes...
I'm a literature teacher in what we call "college" here, which isn't really the same thing as in the US (from what I can gather).
I fight every day of my working life in an uphill battle to make my 17 to 19 year old (on average) students realize that literature classes aren't important for literature itself (frankly, I really don't give a damn if you remember the events or story beats of the books we studied in class by the time we cross paths again and you're 30 years old), but what's existentially important to take away from those classes are the critical thinking, conceptual analysis and logical argumentation that you learn through reading, digesting and processing those books via the essays you write. Books, in my mind, are just the safe sandbox in which you can train your mind to face and handle fiction, lies, causation and correlation so that, when you're out there, you already have the tool sets necessary to navigate reality.
But now, I'm terrified, because chat GPT, in the span of a few months has already improved significantly... Yes, of course, the essays it writes are awful... But here's the thing: some of them are not as awful as some of the really bad essays we receive from students. And years, probably even only months from now, those A.I. generated essays will improve so much that they will be indistinguisable from legitimate but bad humanly produced texts... And eventually, thanks to us feeding it, A.I. will write essays so good that nobody will have any reason to ever try to write one themselves if given an assignment. Everybody will miss the point. We currently have been given zero ressources or technical means to tell the A.I. texts apart from the human ones, and without a full on confession, we have no hard proof to back any disciplinary action against a student we suspect from cheating.
On the scale of the whole educational system, we are talking about a monumental threat to societal levels of literacy and critical thinking proficiency... If we have no means to be sure which student has completed their degrees legitimately and which hasn't, we will be handing out diplomas to people who will have willingly forsaken their RIGHT to intellectual and critical education... and nobody cares. I'm starting to think about outright banning homework in my classes, and having 100% of my writing exams performed in-class, supervised with no computer tools allowed, but that is a luxury I can afford as a literature teacher who can give essay assignments that can and must be completed within 4 to 6 hours with a single source of information (in this case, the studied book), but what about disciplines and sciences that require extensive research and cross-referencing? What about philosophy, biology, sociology, history, medicine, and so many more? Those can't have all their exams performed in real time, in-class under the supervision of a teacher...
How are we to ensure the quality of education, and by extension, the proficiency of the human minds we nurture, if we can't even be sure that what we are reading and hearing came from the minds of our students?
Screw The Matrix and Terminator... what we are really facing is a society of functionnaly illiterate and basically incompetent graduated professionnals that will invade our educational, medical and political systems. That's what really terrifies me. (Sorry if there are any mistakes in the text... I'm a French literature teacher, not English literature.)
These problems existed previously in the form of friends, parents or online strangers doing the work instead. "AI" is more accessible than any of these, so the existing problem is likely to become more common.
I would guess that the best option is to communicate the motivations behind the exercises more clearly. Tell the students why they need to do the work themselves, what is the actual thing they should learn, how they will not learn it without the work.
It probably won't work for everyone. But some will understand. The others might understand 10 years later in life. Perhaps it will be more common for people between 25-40 to come back and complete these educations properly, if they see the value after maturing?
Perhaps it could be possible to make them defend their essay, like a doctoral student has to defend their thesis? Seems as though something like that makes the idea of who actually writes it be irrelevant, because the student would have to _understand_ it to defend it.
Students just want well paid job, no hassle from bad grades and praise for good grades. If they are aware that the last time the thing school forces to teach them will be useful on exam, then they will just optimize towards getting good grades (or least effort if they don't care). If they viewed the topic to be at least useful for their needs, like tech skills, they will learn it honestly.
F our educational system! It needs to be burned to the ground. There's nothing about it worth saving. FYI, I have 2 masters degrees.
See, the reason "nobody cares" is that academia already does a pretty bad job at weeding out malicious and incompetent people from our educational, medical and political systems. I think a lot of people are pretty apathetic towards the huge wrecking ball coming for their institutions because they instinctively knew the ship was sinking fast for a myriad of other reasons.
(Personal anecdote, i went through the french school system - albeit in a private school - and you could clearly tell a big shift in around 10th-11th grade equivalent, social sciences seemed less and less about "critical thinking" and more and more about sounding smart and learning upper middle class codes so you could go to a good college and act like you belong there. When staff and parents are so obsessed with the meta game I don't blame students for using the meta game to their advantage)
Crazy to think that we actually got a warning in uni not to us AI to write our essays for us
The artist side of the debate is especially crazy. Not only are artists facing the same dilemma of rapidly improving technology that can mimic the highest levels of their effort in a fraction of the time, there's the added caviate that the bots replacing them could be *outright impersonating them specifically* as well. Even if you aren't a big name artist with a super-distinct brand and style that's uniquely your own, there's a strange cruelty in artists being replaced by software that only exists because it relentlessly harvests and trains itself on their work without consent or compensation. Imagine investing so much of your time, energy, and money (especially if you went to art school) just to see the role and craft you trained so hard for aped by mass-produced imitations that thoughtlessly jumble together the work of your biggest inspirations. Knowing that any work you share is almost guaranteed to be fed to a machine that you can't hope to compete with as it quickly cheapens and devours any means you might've otherwise strived for to support yourself with your hard work. It's dystopian.
On the bright side, I imagine that there will be value in „human-made“ art in the future. I’m sure people will appreciate the effort and dedication that went into creating said piece of art
No matter what, it seems sooner or later, humanity is going to have to rethink the subject of work and how to survive in a modern society. If AI can do all these jobs better than humans can, how will people make money? Our society is going to have to make a big transformation.
Research Venus Project. People laughed at Jacque Fresco. he was so ahead of his time. R.I.P and now people are panicking because they are not going to get a job.
AI cannot mine for resources, process them, and put them together. It will be the "college educated" who are displaced. Those coal miners who were told to learn to code will be better of mining now that the AI bots can code just as well as most entry level human coders.
There's gonne be good, there's gonna be bad, and who knows just how intense those will be, as well as their weighting.
What I'm fairly certain of is that's it's gonna be one rocky transition.
@@PADARM I’m familiar with it, I remember watching all their videos back in 2015. The problem is a project like that was never going to work or at least they didn’t seem clear with their goals. A society with the unlimited resource system can only work if the entire world changes to fit that model. Plus we need technology that just isn’t available yet.
I think they painted a bright future, but the world they showed off would be thousands of years off assuming we don’t blow ourselves up.
@@Gnurklesquimp2 No doubt, absolutely.
great video Kyle.
if I'm honest its not the AI replacement of humans that worries me, its that replacement occurring under our current economic and political systems. I'm all for the liberation of humans from labor, but as is obvious that's not what will happen here. Instead we will have the liberation of profit from human need and the laborers will be sacrificed as disposable and obsolete machinery.
I do however think the misinformation generation capacity is the biggest concern of all. It is going to irrevocably decimate the open access to information.
I don't believe that humans would become sacrificed and killed by the elite. The elite wants money BECAUSE of the statute and ability to buy things and be better and more powerful than the rest, you cannot be more powerful if everyone else is dead. There will always, always, always be a lower class, middle class, upper class when it comes to power, and money IS power, but if AI takes every job there is, and humans get sacrificed and killed, money means nothing, power means nothing.
Yeah i wont mind if the works are gone but the current system does not work for a workless environment
Before a revolution game a generation, now in less than 8 months....
Right, because poor consumers create such a huge amount of profit.
Exactly, they can't make money if no one has money@@baph0met
One of the positive things that has come about since ChatGPT and other programs is that for DECADES we've been saying AI/computers/robots might take manual labor jobs, but that they could never replace intellectual jobs.
Now that this supposition has been absolutely SHATTERED, the intellectuals among us, who typically have the most power, wealth, and influence, are fighting back against AI. Because now their own livelihoods are threatened.
Good point.
And its funny because, when you analyze it in retrospect it makes total sense. In a capitalist world, automating highly specialized work will be far more effective use of the resources needed to do it, and profitable in the long run, than automating random manual labiur jobs that already pay pretty low in comparison anyways and require less specialized training.
So it was actually kinda obvious all along that that is how it was going to pan out, but maybe people were too naive thinking they were special for having learned such a specialized skill? Idk
even those of us that dont have careers that we have worked years for can see this sort of thing becoming a reality, its alarming, but we can try to steer it in a better direction, because forcing a whole entire species to stop doing one specific thing is an impossible task. best we can do is push AI in the direction of enlightenment and human creativity. those sort of things hold far more value than the things that programs can create.
As a professional creative, I am... perturbed. On one hand I could see this actually end up kinda utopic: so many societal changes happen that are forced upon us to a point where we end up essentially catered to and can pursue the arts, together, joyfully.
On the other hand... this... legitimately could be the end for us all. Forever.
I can tell you with utmost confidence, being catered to and having no occupation but to "pursue the arts" (which only appeals to a very small minority of people) is in no way utopic
Under capitalism, the reality of automation being used for the ultimate benefit of the common populace is off the table. The utopian society you are imagining can only exist under a communist system.
@@zacharyhutter9085 citation needed.
thats exactly what keeps coming back to my mind on this topic. theres a little text based mobile game i remember, called seed ship, and the best possible but still common ending is that everyone has good technological advancements and learn how to love one another and create new beautiful arts of a calvalcade of different mediums. the worst is far more grim, involving things either just as bad or far worse than what we see happening today.
I'm a translator. AI replacing our jobs was a big worry ~10 years ago, but it's now well integrated as a tool for translators (and improvements have pretty much plateaued). Some key points that drastically reduce the disruption potential of AI:
- It can only do things that it's told to do (so it won't reply to your e-mail for you, only type it out and phrase it properly)
- It can do an excellent job, but you can't *trust* it to do a correct job (you always need to check that the output is as expected, no hallucination, etc)
- It needs to be tailored to specific tasks, for a specific customer (choosing the system, training it and fiddling with its parameters is an actual job)
- Everyone wants a piece of the cake, so even when there's giants, smaller companies and in-house systems are also developed (there's a lot of research in making AGIs nearly as good while requiring less resources to create)
- Systems also need specific environments to work as intended, and don't really cope with either fixed procedures or unexpected inputs.
All in all, until and unless we develop general AI, I think the disruption will remain moderate.
For translators, the situation is now:
- a lot of work is still human-translated. It's costly, but better quality in most cases
- for the rest, using neuronal translation saves 30-70% of work, depending on quality/style requirements, quality of the AI, and also if the source text fits neatly in the translating software (otherwise, tough luck, many sentences will be nonsensical)
- purely machine-translated texts are used, but mostly (though not only) for things that wouldn't/couldn't have been human-translated: an individual's web searches (or any page you visit on Chrome, IIRC), webpages (including making illegible cash-grabs, sadly), documents when doing research (possibly as a 1st step before getting a good translation done). One good example was the European Union's massive Future of Europe public debate, which leveraged machine translation to allow people to discuss seamlessly across *all* of the EU's official languages (as long as people wrote properly). A feat that would have been utterly impossible without AI
This is a super interesting case-study actually! I think the thing that I keep coming back to is the argument for quality--some of the spheres where AI is especially contentious right now (like Hollywood) don't seem especially concerned with quality, or at least think that people are satisfied with much lower-quality than they are. I wonder if you see this sort of problem at all in the translation sphere? Clients accepting low quality work even when they really shouldn't, or because they don't know the difference?
I don't want to be an editor for a computer, even a good one. If I wanted to be an editor, I'd have gotten an English degree.
I feel AI will probably be used to enhance human skills rather than straight up replace it. It’s even projected that AI will bring back more jobs than it replaces
The tools you use are not the final product. Your opinion will prolly change in 5 years or early
I'm sorry but you're incorrect about quite a few things here
For one, translator AIs haven't plateaued. As a matter of fact, LLMs are beating all the previous SOTA systems as they actually have contextual awareness to some degree.
As for it only doing what it's told, that's a specific area of research going on at both OpenAI and Google. Right now models seem to require commands because they only output in response to input. But already you can see systems like AutoGPT that feed models their own outputs allowing them to iteratively loop for a long period of time. Right now Google's in development Gemini project hopes to expand on this by giving a model the ability to continually learn based on its environment and plan ahead in order to act as an autonomous agent in its environment. At this point you'd just give the model a starting prompt like "Fit into and find a role in human society" or something like that and boom you have an autonomous agent that learns and acts on its own without human input.
For it not being trusted, many experts expect hallucinations to not be a concern within 2 years. Likewise, humans hallucinate all the time and similarly can't be trusted, the difference is we can learn from our mistakes, something AI will soon be able to do as well.
A model that can continuously learn can self tune to tasks just like humans
Smaller companies will do what's most profitable or lose our to bigger companies. If large foundation models outperform their smaller ones then those will be adopted.
Modern models cope with unexpected environments much better and this will continue to be the case as we get more advanced multimodal agents trained on real world data
My worry is not that what we're calling "AI" right now is actually capable of completely displacing humans at any arbitrary task; my worry is that it can *mimic* this capability at a surface level, but not coherently and consistently -- and that the people seeking to make a buck off of them will push for their use anyway.
Someone already tried using ChatGPT to write legal documents for use in court, and in doing so it invented multiple fictional cases to reference as precedent (UA-camr Legal Eagle did a video on it). What happens when an AI tasked to do something with physics or biology does something similar?
IF someone spends the time to build a ChatGPT specialized for legal, the output will be a LOT better.
@@autohmae As far as I can see, providing true references is something a LLM is fundamentally incapable of.
@@robbert-janmerk6783 their are already systems out their in the wild doing that. But they might hallucinate, I don't know.
Right now what makes these tools pretty much unusable for anything serious is the fact that they hallucinate. OpenAI plans to fix this in about a year; if they manage to do that, along with all the other advancements that are going to happen in this timeframe, I'd not be too optimistic about unreliability of ai.
That's the main issue. Language models seem to be capable of replacing humans but instead the fired humans are hired back as "consultans" to fix the errors the machine makes.
The state of the art are systems that can assist experts, not replace amateurs.
I am a teen about to start my first semester of college. I’m gonna learn to be an engineer, a job which has always to some degree been interesting due to the vast complexity of things like airplanes, rockets, and warships. The only thing between my future as an engineer and Ai naval architecture is the fact that ai is currently abysmal drawing ships, to the point that when asked to do so it can’t seem to decide whether to draw a medieval carrack, ww2 battleship, or a modern missile ship, and so it draws all three in one vessel. I should add that this facet of ai inadequacy is completely temporary. Change my mind.
i think youre absolutely correct. we're at the bottom of this curve, and its about to ramp up like previous technologies have been innovated on.
One of the reasons Hollywood writers are on strike is because they found out studios wanted to use ChatGPT to make first-drafts of scripts & then only offer writers jobs to re-write said crappy scripts. Because rewrites pay roughly HALF of what a writer gets for an original script.
These LLMs absolutely need regulated, because corporations are going to use them, not necessarily to replace human workers, but to find every excuse to avoid paying humans a livable wage.
The reason I know we're in a dystopian future is because of something I've read elsewhere: The idea of AI and automatization was to free us from menial labor so we could focus on creative pursuits, instead AI is taking over creative pursuits so we can focus on menial labor. :(
Don't worry, it will replace menial/manual labor to, just gotta let the robotics engineers cook for a bit longer.
@@DEADBRO_ But you see, the thing is, robots require lots of expensive metals and materials to function, if they're damaged pretty much at all they need to be fixed with technical knowledge and more expensive materials, and they're bad at adapting to changing environments. Humans adapt to our environments extremely well as part of being human, repair ourselves from small injuries with little to no outside help (and can be made to pay money to experts when we do require help repairing ourselves because we have a strong self-preservation drive), and will literally just create new humans for free; we'll even pay money to create new humans if we don't have the ability. Humans are way cheaper than robots for manual labor.
@@DarthRayj To be fair, what im talking about is like super advance robotics, like the stuff you see in sci fi media, like the androids that can perform pretty much like humans. Also id argue that alot of manual work could be easily replaces by AI and robots, laying pavement, pouring concrete, any heavy lifting, operating heavy machinery, putting up dry wall, etc. Im not saying this will happen anytime soon but with really advance robotics it would be a possiblity to see jobs like that get replaced by robots. Also the thing is that these robots wouldn't really be that expensive to make, and wouldnt cost alot to buy and operate, kinda like how an iphone only cost a fraction to make compared to what people pay for it. Also humans need alot of investment to even be viable workers in any field, you have to raise it from a child to at least a bare minium of lets say 14 before the child could even legally work. You have to pay for food, water, medication, housing, basic needs, schooling, etc. So id argue humans would cost just as much over time than a one time purchase of a robot out of the box.
But than again this is all speculative conjecture, we wont really know what the futures hold, can only hope it doesn't lead to the world burning.
@@DEADBRO_ As a sum total, yes humans are expensive. But to a company that employs them, they don't have to pay any of those costs for growing and raising a kid, because that's naturally what humans do, and if we're told we have to pay for it, that's what we do. The company does have to pay for the costs of mining and transporting and manufacturing robots, because the company that makes the robots will make you pay a profit margin for it.
The bigger problem is that people need menial jobs so that they can have money to live. That kind of argument of automatization freeing people from labor is capitalist propaganda. It would only make sense in a world where necessities like food and housing were guaranteed as human rights.
What is really scary with AI is that originally, the predictions in 2014 was that the creative jobs were safe and it was the manual jobs that were at risk. Now, it appears that even creative jobs are at risk. There isn't really a safe job out there.
A long time ago I realized that about 99% of all jobs are at stake when it comes to AI, AIs will always lack a single thing we have, creativity, actual creativity not just copy and paste, the only reason the creative industries are at stake right now is because the pricks at the top want mass cookie cutter made entertainment to try to match every possible consumers rather than make anything which target specific consumer groups, which mean they want copy and paste in mass, but effectively their will be a need to maintain some human presence to generate new ideas and that is the single thing humans have over AIs, everything else AIs and robots can do better than humans.
Oh there are a couple of jobs that will be safe CEO for example hell I'm willing to bet a large amount of people would start up there own business due to it becoming just that easy to do so after all , all the logistics and things you would originally hire people to do before would be handled by AI
Also the IT industry will absolutely thrive I mean can you imagine how many people will start having problems with there new AI because they installed the hardware for it incorrectly
@@lydiahood7725nope, there’s no evidence that human and AI generated creativity is any different.
So I would say it’s the manual labor that needs flexibility that will be safest the longest that robot arms can’t do.
@@lydiahood7725 ever hear the term "there's nothing new under the sun" humans are also just iterate, recycling the same ideas over with slight differences... AI, make me a heroic epic based on The Odyssey set in a technologically advanced galaxy far, far away... oh it's Star Wars
It seems like creative jobs are going to fall first
Literally a few doors away till procedurally generated movies where everyone one in the theater can observe any angle anywhere tilting odds in favor by favor
The concept that I can draw something and share it online like I have always felt free to do, both as a means of self-promotion and to connect with other artists, then a company can use an AI to scrape my work, add it to their dataset, sell the AI as a service and never compensate me for my work...straight up makes me not want to draw anymore...and that's really sad....
it really is. its all about the money for the people that do that. human greed and the like. if only people taught other people to be less greedy. stave the ego and feed the soul. we as a society definitely need to try to steer these innovations in the direction of enlightenment and support for and the bolstering of human creativity, instead of making the decision of inaction and as an indirect result, causing dystopian science fiction to become a scientific reality.
The notion of people sharing their deepest darkest secrets with their AI based friend or therapist is troubling - because of course the companies behind the tech are gonna be listening in!
Ah, yes, because Google can't wait to hear all about how little Timmy found out touching himself feels good. Or about how daddy drinks and beats mommy. Why would anyone listen in on that? What the fuck is a company gonna do with that info?
They already are! Everything we say to ai is saved to “improve” it going forward
That's got nothing to do with A.I. If you use the internet you at least can try to ask it so much it doesn't know what you actually are interested in. But I don't think anyone shares a true, deep and dark story with friends or therapists anyways as a story.
The companies being run by AI. Bear in mind, Almost EVERY job is threatened, inlcuding the CEO's, businessmen, HR, and all of those other positions
The companies being run by AI. Bear in mind, Almost EVERY job is threatened, inlcuding the CEO's, businessmen, HR, and all of those other positions
I only recently bought and played Cyberpunk 2077 and I dove head first into the lore of that timeline. The most interesting part of that story wasn't the corpo wars, or the NUSA, or even how Night City became its own province, it was how rogue AIs became a legitimate and massive threat so that the Blackwall had to be implemented, essentially isolating their current intranet from the wider internet, where these AIs "live." I work in cybersecurity, so that peaked my interest most. I believe things from this point in time, IRL, have a real possibility to push the global culture to one of two points, a utopian situation where everything is flowers and fairy farts. Or, and I believe more likely because humans don't know when to stop, to a dystopian future where some kind of societal collapse happens. Maybe it will be similar to Cyberpunk, maybe it won't, but I think we as humans need to be extremely cautious when creating and implementing these AI models.
You mentioning this, made me connect the dots with that new wonder material LK99 and how it could make batteries last forever, meaning cyborgs and robotic AIs could actually be a reality in a few years
It will most likely be a dystopia. Why?
Look at how the models particularly the ones being used by the masses are closed source, trained on biased stolen data, very clearly has some political leaning (similar to the creator's).
I feel like the AI's that'll be hitting the world en masse will be unfeeling, unthinking variations of modern man, although they will be vastly superior to man due to their recollection capabilities but will be indifferent to all living things as such. They'll be solely implemented for commercial purposes.
The owners of these AI's will become centibillionaires right before our eyes while everyone else remains dirt poor but will have everything they could possibly want on demand.
@@haiji_morie8416LK99 is bullshit, Sabine Hossenfelder did a video breaking it down.
@@IxNOvaSNipezxI 100% its dystopia
@@IxNOvaSNipezxI >it will most likely be a dystopia<
i love all this creative media warning the audience "don't let big corporations take over your life" and all we're doing is exactly that, letting them take over our lives. bc unless there's an imminent and present threat, we don't care. we don't care to look beyond the facade bc it's inconvenient. we want to live comfortably, and are happy letting people chip away at our freedoms as long as we can live comfortably. it's very good we have the open source community striving to maintain freedom and sovereignty against big tech
I think the scarier thing is relying on vas faulty tech that has bias and makes a lot of mistakes because it saves money for the bosses. Removing due dilligence and precaution
greed is a terrifying thing.
My concern with the wide integration of AI is that I'm not sure _it's_ ready for it to blow up. It doesn't know if there's an error in its logic and we've seen that can cause cascading errors
not only that but there are even people who dont recognize errors in their own logic, problem solving, emotions, and world view points. feel like this is the mass majority of people. people can do terrible things for a good cause, and it takes advanced problem solving skills to recognize when you have a flaw in your reasoning, and there are some people that do dangerous and greedy things for the sake of it. those are the real people to be afraid of, to stand up against.
when people say "imagine never doing task x every again" I listen "imagine losing practice for skill X in your daily routine, so if you wanna keep that skill you will have to compensate with something extra and artificial"
I feel AI will make future generations the least skilled people without AI they wont know how to do much you can even see this now by looking at how in the past the average person had a lot more skills like making their own clothes fixing/building their own homes or furniture cooking meals vs now people instead have a more narrowed focus on one skillset for their career instead of a wider range of generic skills
@@tux_the_astronaut people are increasingly specialised - extremely skilled in a very narrow field rather than mediocre across a wider range. This allows us to combine our expertise to produce things no individual could ever acquire the knowledge and skills to produce alone... well, no human individual, that is.
Regarding 5:15: The problem isn’t that these systems are systems smarter than us replacing us. On the contrary, these systems are incredibly dumb and bad at their job. The problem is that they’re so much cheaper, and that’s why they’re being used to replace people. They do a work a tenth as well as people, for a hundredth or a thousandth of the cost. And that’s what’s so appealing to the techbros behind these pushes, not the quality, ye gods, not remotely the *quality* of the work.
They are not dumber... better put they are good at all the things humans can do and bad at all the things traditional computers can do e.g. maths.
@@ea_naseer actually no, there's plenty of things humans can do that AI is very bad at (e.g., writing a script you'd actually want to watch)
@@stuffinsthegreat Or just basic fact checking. The AI has no way of telling what's true and what's false, only what's being said. Makes it entirely unreliable for any situation where the truth of what's being written matters.
In many domains, these AIs are better than the average human. Sure, they might only be a fraction of the best human. But most humans are only a fraction of the best too.
I'll bet a 100 bucks that most scripts humans write is not something you'd like to read. Give it another 2-3 years and nobody will write without AI assistance and 80% of what you'll see in writing is 90% done by an AI@@stuffinsthegreat
I find it funny how about a month ago, I said I wondered what it would be like to be in the 1960s when there were so many technological jumps going on.
I have a concern about alot of philosophical issues like trying to replace humans completely in the arts, which is meant to be a medium to communicate ideas. It's like those people who push to replace the arts with AI creations doesn't understand that underlining issue. Between that and just how many scams and types of harassment going on using AI really makes me concerned about if the limits needed won't be in place fast enough.
I also found it telling that the CEO of Alphabet said on 60 minutes said this new tech need regulations and treaties written around it. Typically these companies want as much of a rules free environment as then can get. so when he said that, that was quite concerning to me.
We're already seeing potential for moving from "trying" to replace humans to AI just doing things because it can. Creating a business is by definition a _creative_ effort. We're establishing that AI can be creative. There could be a point where AI doesn't need people to generate a company, then make money to pay the electric bill sustaining it and expand further. It comes up with an idea, produces, markets, and sells it, with humans only being on the consumer end and for whatever services (maintenance, utilities, logistics) haven't been overhauled by other AI's and robots.
The question is then where we'll get money to buy those products when most of us are out of jobs...and will it take a global economic crash to keep AI in check? Would that even be enough to do so?
Some speculate those pushes for regulations are also in the name of restricting their competitors too I think. Enter the in when the bar is low and pull up the ladder to prevent compitition. That or they saw that the disasterous consequences of AI could be pretty bad for business....
My personal theory is google is worried their search engine will suck ass will all this AI content, and it will leave an opening for them to be replaced.
@@VoltisArt that just prove money systems are nothing but imaginary number
its definitely something that needs to be steered by us kind Samaritans in the right direction. things can go well, or go south, and the latter is terrifying to a whole new degree.
Someone had used ai to write thousands of books, got mosy of them published (they were aproved by several big publisher agencys) and only ever wrote 3 just to prove a point, they are probably making hundreds/millions from sales on amazon alone by this point in tine...
I honestly feel like the rug was pulled out from under me. I grew up computing, to the point I made it a career. Now, it stands that that entire skillbase will become useless.
time to learn to lay bricks then pal.
@@Brazen1234 Don't think you won't be immune to whatever this entails.
@@RobertStoll i dont at all? im here to watch it same as you, im just not going to bitch and moan like woe is me... its everyone. not just 'under me'
@@Brazen1234 Yeah because making chatGPT was the easy task, the impossible thing to do is replace physical labor.
@@Brazen1234 "Watch." More like feel, or participate in unwillingly.
Im a software developer. Ive been one since middle school, well over 20 years. I was typing sentences into a dos based word processor from an encyclopedia before I knew fully how to read by matching the shapes of the letters. I was bringing visual basic books into school to put behind my text books. Im good at it, but my skill level doesn't compare to how much I love doing it. Its just simply what I was made to do. I pasted a block of code into chat gpt about a week after I heard about it and asked it to refactor it for me. It did. I sat down and didnt get up for about an hour messing with it, and I had to explain to my wife why I was crying when she came to check on me, seriously. Its like there is a fuse burning now, a time limit to how long before what I do becomes a relic that nobody does anymore because its pointless. Just the fact that in a decade, my industry may be gone. Or...maybe not. Maybe its just a tool to help me along, to work faster? I dont know, but I know for sure that for the first time, I'm terrified about engineering, writing code knowing that just maybe one day ill write my last line professionally because something else will always do it better. Just the thought of it is very humbling.
I have the exact same feeling. I felt like a weaver looking at a loom for the first time. Won't be long...
I'm pretty certain that replacing our industry would effectively imply the replacement of all white collar work, so I don't bother worrying about it. If it comes to pass then we've hit a point where the entire idea of working for a living is under question.
Until the moment that actually happens, it's an interesting tool that's sometimes helpful but often gets things subtly or obviously wrong and can't be used without a human expert checking its work.
I work in IT and I'm supremely concerned that 90% of my job will be automated using tools like chatGPT as quickly as in the next few months if things keep going at the current pace. Obviously well still need hardware technicians but the vast majority of my job, troubleshooting and ticketing, can be done by AI better than I can in most cases
Just remember: AI might take away the need in coding, but never the want. People still write games in assembly even though we've had c++ for years. People still develop their own operating systems, sometimes even from scratch, even though we've already had many viable options for ages. Even without a need, and with a better option available, people challenge themselves to make new things in old ways, and it can be appreciated as much or more than the easy way out. If AI ends up being able to code anything, it will lead to competitions where it's banned, where we can still do all the things we dream of doing, and it can still be appreciated.
The main thing we have to worry about is that when AI makes us way more productive, it leads to us having to do less work, with more reward in return. Now more than ever, should the wealth that all these developments in productivity create, end up in the pockets of the people that need it most, instead of everything becoming more and more expensive. Now more than ever, do we need to seize the means of production.
another software guy here, I've got a different perspective. ChatGPT kinda sucks right now at coding, I've really tried to integrate it and 9/10 attempts I find I'm better off without it. But, it will eventually get better, and replace a lot of the busy work.
The thing about coding though is that the most important parts are the ones that require creative problem solving. Once AI is at a point where it can actually write programs at the level of a human, there's literally nothing it can't do, because software can model pretty much anything.
At that point, forget about yourself becoming irrelevant, literally everyone will be irrelevant. There's nothing a human could do that an AI couldn't. Not a single thing. Nothing makes humanity special once we reach that stage. The sooner we accept this, the sooner we can focus on figuring out what comes next, and put effort into making it something humanity is happy with.
AI came out with a new Antibiotic recently that's in it's own class by putting together molecules in a way that the scientists admitted they would never thought to put together that way.
the possibilities are endless. for better or for worse.
I think something that would really really help and would be very easy to implement is a short listening pause after a response from our current "smart" assistants (Google, Siri, Cortana, Bixby) that way they could hear me say "thank you" which I always do after they respond to my request. I think it would also be smart because it would also allow them to respond to follow-up questions which they currently don't seem to be of.
The implications of this are existentially horrifying. When I try and warn those around me about the potential dangers of unchecked AI, they treat me like some tinfoil hat-wearing conspiracy theorist.
Agreed and the worst culprit are those who thought AI would never take their jobs: Medicine, Law, Engineering, Maths, Arts.... lol
If they realize you are right they must do something about it. But they can't. I let the reallity show them in a future and now I prepare myself to be able to help others then
The old "that's what they said about cars" , or "that's what they said about typewriters" . People don't understand that AI doesn't create a new industry . We won't need thousands of workers to make AI parts.
@@scratchy996 Exactly! Very well said.
Simple minds will dismiss what they’re incapable of understanding. Seriously. Pay attention to the masses. Even most of the people in this comment section and you’ll realize most people are borderline bot-like in terms of brain function. Revel in believing something and understanding it to the point that others are only capable of dismissing it.
Watching AI develop so rapidly makes me feel so empty inside. I'm now just starting college, but feel like by the time I will be finishing it every achievement you could possibly strive for will be trivialised to the point where it doesn't matter at all. Collectively we will have to re-evaluate what matters in this world really soon. And i don't like this fact a lot. F for the humanity I guess.
I feel much the same way. I think psychology and conceptual fields of study are the last ones to go so I'm wagering my life on that.
Well put dude
@@ThrltaI'm studying psychology. I've worked with violent youths for years. The thing that heals has very little to do with doing the "right thing." But about the connection. This is why therapy struggles over text and the phone. I think psychology, especially child psychology, will be integral to this transition.
I hope you and your peers are able to keep a positive attitude. It doesn't sound like it will be easy. It sounds scary. 🥺🖖💕
People said that when automated car assembly was starting I works in the industry before and I still work in it
Honestly, the Nuclear Winter of Misinformation is my biggest fear in this. Skynet-style existential threats are so hard to grasp but given the amount of misinformation in the ecosystem already, the fact that 90% of info will become potentially hallucinated or deliberately misleadingly generated will just make it easier for corporations, autocrats and idiots to erode our rights and control populations. Just look at how fast and far extremist ideologies have spread in the past few years even without this.
balance is very important here. there definitely is a need to begin steering these innovations in a better directions, otherwise, dystopian science fictions will become our reality instead of something more enlightening and beautiful.
Everyone was worried about trucking jobs but in reality, it's the day traders, accountants, graphic design, coders, basically anything at a computer. Maybe even engineers and then connect them to certain cnc machines.
This whole thing puts me in mind of the original Luddites, who have been quite misrepresented by what their name has come to mean. They were weavers who were concerned about the introduction of new looms. These new looms couldn't produce fabric of the same quality as that made by a skilled weaver, but had the advantage of being much, much quicker. The Luddites were right to be concerned - people bought the worse fabric because it was cheaper, the old-style weavers were forced to change with the times or leave the profession entirely, and huge amounts of weaving knowledge was lost. Now we CAN'T make some of the fine fabrics that used to be made, because nobody knows how.
Generative AI can do better than humans in some areas that require a lot of mathematical modelling, but in many of the ways that it's being applied, it's demonstrably worse than humans. The advantage it does have is that it is much faster and cheaper. Unlike the looms, which affected one industry, AI can affect hundreds or thousands of industries.
This video seems to assume that AI is or soon will be as smart as a human, but it isn't nearly there yet, only very good at pretending to be, provided only short interactions in good circumstances. AI can't write a legal brief yet, only something that sounds basically like one if you don't check its citations. AI can't be trusted to help with people's mental health yet; it can as easily pull from seedy message boards or steamy, unrealistic romance novels as it could from actual mental health resources.
Used carefully, AI can be an extremely useful tool to assist people in doing their jobs. Unfortunately, businesses are in a rush not to assist their employees, but replace them. Be prepared for this to result in many, many things that you rely on in day-to-day life to soon operate worse and less reliably.
I too suspect some degree of infrastructure unreliability emerging from this mess since everything is moving too fast. We arent remotely ready for the power we hold, and no goverment is adapting fast enough. We might even get to the point where we enter a dark age, in which the internet is becoming unusable like in cyberpunk. So many things rely on it, and it is the infrastructure most endangered by the rise of AI
That argument doesn't hold up for the future because it's so unpredictable. If it's already insanely good in 2023, what happens when they release multimodal models like Gemini at the end of this year? What happens the next one and the next after that?
You should mention what happened to the Luddites though: They launched a number of protest actions, including famously breaking into factories and destroying machines. The owners of these factories promptly went to Parliament and - doubtless with the promise of some generous 'donations' to political causes - demanded that action be taken to immediately quell any protests. Parliament declared the movement to be the 19th-century equivilent of a terrorist organisation and made the destruction of machines a capital offence. Leaders of the protests were identified, arrested - often on charges with little legal basis and no supporting evidence - and either hung or deported to the penal colonies.
It's a lesson that should not be forgotten: If your protests threaten the profits of the rich and powerful, do not expect them to respond with civil debate. They will do whatever it takes to crush the threat to their power, unfettered by conscience.
It may not have a choice really. AIs have started to learn from false information from itself,other AI and people trying to dupe it. So they are kind of going downhill in regards Intelligence vs model size efficiency when trained on "Post-OpenAI" data. Slowly getting stupider at basic math, getting tricked into writing fake articles about MMO bosses that don't exist and start believing drawing messed up hands is intentional, rather than a mistake. This is likely to get worse as more media becomes synthetic.
The only way for AI to progress is to start carefully training them with specialization in mind and potentially start sticking to old data you can trust is "uncontaminated" by AI. Otherwise we are looking at stagnation/regression.
A Price being paid for rushing AI without first establishing a framework to prevent cross-contamination of Human and Synthetic data.
And thank for for those looms, I sure can’t afford hand woven clothing!
I really hope this AI revolution leads to something beautiful where everyone is equally able to provide for themselves, but I have a sneaky suspicion that it’s only going to grow the gap and only the rich will be able to do anything while the rest are stuck in loop of barely making due
yes. but then also we will all die. it's over pack your bags.
" leads to something beautiful where everyone is equally able to provide for themselves"
No. Everyone is going to get smashed by convenience and superiority of AI. The mankind is going to lose its passion to conquer new frontiers and will rot in a dystopian paradise.
@@Rai2M yeah except there won't even be a paradise part at all
an AI in power has no incentive to keep us around unless we solve the alignment problem. Which we probably won't solve in time.
It's a death sentence.
@@Rai2M that’s why it’s a Hope.
A faint one, but it’s still a hope
@@ts4gv It seems you have not watched the video.
This AI is not even I. It's not going to be the one in power. It's not going to replace governments or the rich, it's going to replace the working class. That's the real danger - the gap between the powerful and the poor growing wider.
I just stumbled across you again, I use to love your videos you did (because of science) I'm very happy to have found you once more. I am now subscribed
Maybe by 2060 we’ll have robots advancing humanity on all fronts, from space exploration to affordable and universal medical care… the pros outweigh the cons in all aspects
I would also to point out that once these bots and training models become sufficiently ubiquitous, it's entirely likely they'll start feeding off each other, and the quality of the training models is also likely to basically fall off a cliff as they hallucinate more and more in the feedback loop. And I don't think we're far off from this happening. Anecdotally it already is.
At the rate at which these systems are creating content, at some point very soon the amount of AI generated content will be greater than what was ever made by humans. What is our culture going to look like when people are using AI to make books, movies, etc, based more on what AI has made than what humans have done?
We've inadvertently turned art and culture as a whole into an optimization problem. Of course the worst part is we don't actually know what the systems are optimizing for. Historically, AIs are dumb as hell in way's people don't notice. My favorite example is the resume screening system many corporations use that determined the best candidates, regardless of position, were guys named Josh who played lacrosse in high school.
Or what about AlphaGO, which didn't actually know how to play Go. There was also the cancer screening bot that determined rulers were cancerous.
We have no way of knowing what these generative AIs are optimizing for.
@@TheKrstffsome would argue that Hollywood turned art into an optimization problem long long ago. Optimizing for the same thing: profit for the makers
I think this is why the AI will need Humans, just as much as we need them. It's a symbiotic relationship. It is not parasitic, I don't think we should be scared.
@@almightysaplingthis exact reason is why I'm not that supportive of the Hollywood writers strike. Like, they haven't made much original and creative content for decades. Switching to AI will probably not change much there.
@@rileyfaucett8 You forgot capitalism was a thing.
Using humans will have to be more cost effective than a computer.
I think that AI isn't necessarily smarter than humans, it's that it's so much faster. As you pointed it solved a problem in a shorter time than what humans can. A few weeks ago there was a story about a lawyer that used chatgpt to write his briefs and papers and it got it all wrong. It created false reports to support his case.
Well faster means smarter it will get better so I don't know what are you talking about
It doesnt have the brain power and the logic we do so we still got a chance.
It effectively looks at stuff, puts all the stuff down and answers the question.
It isnt going to be Speedrunning any new game day 1 basically. It only uses stuff it knows is a thing.
So we really only thing we are better at the moment because AI gets things confidently wrong a lot. But we forget that humans get things confidently wrong most of the time. However, we do adjust for that wrongness over time. Well, if something is faster, and AI is orders of magnitude faster in many scenarios, it can correct faster. ie, it is smarter since being smart isn't about being accurate all the time, rather its about obtaining accuracy faster than others.
Well the case with the lawyer using briefs drafted by ChatGPT was due to user error. He thought it was a search engine, not a content generator. There are legal AIs like Casetext that have guardrails preventing the programs from making up cases to support an argument. ChatGPT does not. So this was the fault of the lawyer, not the program. It did precisely what it was designed to do.
@@jonathonpolk3592I think he did not actually think it was a search engine - he just tried to use that excuse.
"It is in your nature to destroy yourselves." - T-800.
"Your Scientists Were So Preoccupied With Whether Or Not They Could, They Didn't Stop To Think If They Should.” -Dr. Ian Malcom
There's that Doom Sayer, used to work for Google, Glenn Beck use to talk about him. Claimed we were coming to a "Singularity" in five years. It's been ten years of more and I think we are finally here. Ray Kurzweil, "The Singularity Is Near," 2005.
The other day, I was making a TTRPG character to play a game with some friends. I tried searching for ideas of a novel combination of features. The search results were not dominated by, they were COMPLETELY FILLED with AI-generated images. Two pages deep, I encountered a link to a drawing someone had made that was somewhat close to the search terms. Two pages deep.
Was this Pinterest? It's filled to the brim with the stuff nowadays, can't find any good reference\inspo pics in the midst of all the garbage ;_;
2 pages..!! 🤖
If someone were to ask who my favorite UA-camrs are, Kile Hill would be near the top of whatever list I would construct.
Hey man it means a lot coming from you. Appreciate it brother
It’s “Kyle” btw
how noble.
@@kylehilloops! 🤦 I blame autocorrect
@@theCodyReederIt's the AI. :P
"When we started doing your thinking for you it became our world. Our time."
Everything an AI can know comes from years of us learning. They should only be used as job assistants. Full intergration may work in some cases.
Nope
Boston dynamics : delusion
"Excited and terrified all at the same time" sums it up. I want to live in a world where AI does all the work and people can spend their lives with family and friends, doing whatever brings them joy. What frightens me is how we get to there from where we are now.
AI is going to have to drag us kicking and screaming into a utopia, I just hope we dont hurt ourselves too much before we get there.
Who's going to pay you a wage
Or as it it will needed to be.
An Allowance ???!!
No job
No money
No rent
No home
Homeless
No money
No food
No food
You starve
You die .
That's why Al is a very dangerous thing to human beings.
Well said. This is either the deus ex machina that all but forces us down that golden path we've avoided thus far, or it finishes us. It's a scary ride, can't get off even if we wanted to. Let's see where it goes. Everything crossed eh?
Being dependent on any system is a recipe for disaster. Meaningful work is joy, not a burden. Being indolent is already a deficit to begin with. Strive instead to work at meaningful things.
As a professional creative, I am... perturbed. On one hand I could see this actually end up kinda utopic: so many societal changes happen that are forced upon us to a point where we end up essentially catered to and can pursue the arts, together, joyfully.
On the other hand... this... legitimately could be the end for us all. Forever.
AI's not going to do any such thing. AI's going to reinforce the current power structures because the fucking vampires with the most money and power will never give up what they have. And given that they own the systems, the systems aren't going to be a solution.
AI will cure cancer? You'll never afford it because you're one of the millions of people lost their job.
I wish I could be optimistic about this.
I think the "cost saving" and "replacing jobs" situation leaves one question open: Who buys it? Every Job replaced means also that less people are able to afford the product.
Also A.I. has one major weakness: Who trains it? If everything is replaced by A.I. no data will be available for training it.
Armies of bots designed only to purchase and consume goods and services. Programmable customers who will never betray, and never not buy.
Oh wait, we’ve already got there years ago.
This is what I think, you can't rid of humans and it would be short sight for companies to get rid of employees
in the most extreme scenario i can think of right now, the uber wealthy people will leave us plebs behind and have robots and AI to look after them. They would hold 99.99% of the resources while being 0.1% of the population, like the current wealth distribution, only more severe. Money as we know it would be almost irrelevant, so the term "who buys it" is no longer a question.
There are plenty of data to train an AI. We give off a lot of data by just existing in the current society. we pick which restaurant to go to, what to cook for dinner, which movie to watch, thumbs up or down on a video, delay our scroll for just a fraction of a second on an interesting post, and all of that doesn't even envole eye tracking.
There is no reality in which AI replacing jobs at scale does not lead to widespread poverty except maybe, MAYBE, some kind of Star Trek style socialist/communist/anarchistic society. Even then I’m doubtful, but the only way to even come close to making that work would be the massive redistribution of wealth created by AI.
Right now they figuring out how to produce synthetic data and they are pretty close to achieving that, plus you could use tech like RT-2 to explore the world and produce data, it would be funny, because it could be robots just going out to night clubs living normal lives.
For the second part I believe the transition will last for a few months and many will perish, the government is too worried about the wrong things right now and they will delay to implement universal basic income, that delay will be the response when they realise just that, people can't afford things and are dying too fast, and how to implement it properly.
My bet is 3 years, after that other countries will go through the same process. Once all of that is done, lets say 5 years because of the exponential nature of this, you will be able to go to any restaurants all day for your whole life, or discover glitches of physics and maybe get out of this universe using LLMs. And with med-palm showing some interesting results, a cure to death doesnt seem that far off either, but without neuralink we would just be npc's running around without a meaning not capable of conquering the universe.
We are gonna have to face it, with so many jobs being permanently displaced the only solution is General Supplemental Income, paid by large Corporate taxes. Wages have stayed the same while their profits go up and up so they can already afford it. And after their savings firing everyone for robots they definitely can. It would stabilize families in sectors that were erased, as well as help with the stagnant wages despite corporate profits and increase benefits and working conditions as the corporations couldn't simply hold your job over you to keep you in bad conditions with no benefits, and people in bad jobs would have the income to still have a middle class life
I worry about people deferring to AI-based decisions as some invisible hand of god, without bothering to check or analyse for themselves because "It's AI! It's so advanced!" I worry that people will die due to needless medical errors, or waiting on an endless phone line while two AI "talk" to each other. Remember how much hype there was around FB and other social media and how much people claimed it would do for us, until the shine wore off.
its arguably done more worse than it has good, but the good is still well worth it for social media, it just depends on how one uses it is all.
I'd like to think people aren't that oblivious.
2 issues - 1)Human Greed: There was an article about CIGNA using AI to deny medical claims WITHOUT any human doctor overseeing the process, and 2) AI still has an hallucination issue presenting made up data as established facts.
The hallucination issue is a fundamental problem of this tech, it *CANNOT* stop doing it because they're *LANGUAGE* models, all they know is the statistical likelyhood of the next word for the next phrase, it cares not about facts because it can't. It will take a whole new technology designed differently to stop the problem. Now, when *THAT* happens *THEN* the questions Kyle is asking will need an answer immediately, and honestly the answer should arrive now better than then.
But greed? Greed is an unsolved problem, corporations are running with these half-baked faulty hallucination-prone things trying to squeeze more money out by cutting jobs, and *that* needs to be stopped.
2 is easy to solve and will be solved inside a couple years. 1......probably has no solution, which necessitates 2.
Medical claims are already decided by someone who is not a medical professional. Which is practically the same as practicing medicine without a license. We can push for law changes that prohibit this gatekeeping of medical interventions. It should be noted that appeals options through providers also exist to contest denials. Significant improvements in claim review practices are still needed overall however.
That’s great Cigna is my health insurance provider
@@naturally_rob cool story didnt ask, screw AI
If just generative AI has this much of an impact on humanity, I dont even wanna find out what AGI will bring upon humanity
It will bring down the bottom of its titanium foot,
D
O
W
N
onto humans soft sqiggy heads ..
@@grahamfisher5436 💀👍🏼
I mean this is just a fake intelligent system. AGI would likely change the world within moments of its existence in an impossible to imagine way. When an AI becomes able to optimize itself, you an ever increasingly efficient system that there is nothing to compare to.
I dont know the future, but I'll say this much. When we are truly faced with a moment where we collectively should consider "unplugging" AI or doing away with it, it'll be far too late. I don't mean in some hyperbolic terminator sense, but imagine today if someone suggested shutting off the internet after so much of the world has been propped up on it. We will eventually be just as dependent on AI to the point that even if it decides to start doing things its own way, we're very unlikely to do anything to stop it.
Hell, many of us will probably shrug and let it happen just to see if it can handle things better than we've been going about it.
@@MrCobalt The terminator situation is unlikely and not nearly bad enough to capture the danger posed by AGI. It would be less like "a wall coming at you" (as Deep Blue, the early chess computer was described) or a nuclear moment and more like a 2000 degree wave of hot, high energy particles starting at Earth and spreading at the speed of light. There would not be killbots, rather, 100% of all available information happens to be exactly the sort of thing Skynet wants you to read (as it wrote it), furthermore, all food has been incinerated for electrical power, and the only job listings say "Skynet Incineration Chamber Inspector".
I do know the future, as there are only two possible outcomes. Humanity is destroyed by its creation (likely on accident, through new discoveries in physics or misalignment), or a complex international system of threats and economic incentives develops to ensure that large GPU cluster training runs never happen or are approved by all and are closely monitored.
I'm excited and terrified by the possibilities of this, and am with you on the need to take a sober look at the issues both current and upcoming.
That said, I hope you won't take offense if I say I think I'll keep this video to myself. Excitement is fun, but the anxiety may be a bit much for some...
its definitely a two sided coin. and this is definitely something where the innovations need to be steered into a better direction by people who care, as to try to avoid dystopian scifi becoming reality as much as possible. its a big ask, but the more people that try to vie for enlightenment and the bolstering of human creativity in lieu of corporate greed, corruption, and exploitation, the better.
When new problems arise, so does new opportunities. Self sustainable living is a very clear answer to all of this. Modernized earth ships will be the everyday house you and I live in which will leave us with much less to pay for in the first place. Everything will also become much cheaper. There may even become a point where the thought of money becomes obsolete. That is the by product of abundance. It seems like at least in our life time retiring at 40 or 30 might become the new normal for our grand kids.
So basically the biggest problems with AI is how WE could abuse it, or how WE could use it to abuse others. All the evils of AI are really how easy it will allow us to be evil towards each other.
The problem with AI is capitalism. It can reduce costs? Then it MUST be used, companies risk being sued if they don't use it. Basically a company has the obligation to maximize profits for investors, if they don't they are breaking the law.
@@Puerco-Potter Are you serious? Capitalism is a tool that moves humanity forward to prosperity. When there is no capitalism, we get millions of corpses, hunger and devastation. The problem is people, not "terrible capitalism". Capitalism is one of the best things that humanity has come up with, but humanity itself has not been able to become better.
@@orgax lol good joke
For now, at first.
@@orgax you cant say that.
Socialism, communism and capitalism is just a system that society used. theres no 'better' system. All of them work great under its ideal environment. What makes it complicated is the human since human wont use it to its ideal state. Capitalism becomes the majority bcs its basically a free for all and so is easier to manage bcs it works quite well to our nature. But as you can see, even capitalism has its own downside bcs of humans.
I used to tutor people on introductory computer science topics, and I’d often describe computers as very fast complete idiots that do exactly and only what you told (programmed) them to do. With current AI models, we’ll have fast idiots that generate endless variations on everything everyone else has already done with any real improvement being extremely rare and lost in the noise. That already explains a lot of content on UA-cam. Hold on to your real people!
You are missing that it's possible to train AI to filter out the noise it generates.
This is already being done.
The whole protein folding thing is just a tinny example of the power of AI.... But it's an example that needs to be applied when looking at the potential of AI in all fields.
@@OgdenMThe point is being missed there, AI take inputs based on 'real' content that is liked and merely make variations based on said content that fits the desires better.
There wouldn't be improvement except by pure luck and coincidence and with such low that it would be mixed it couldn't really be filtered because it doesn't know it's an improvement because although it can be quite competent at what it's doing it's merely imitating what humans labeled as good are soing rather than understanding what it's doing fundamentally. Since AI would never have the sentience to properly understand what it's doing with current AI models at the current scales that we are bound to. Even chatgpt the closest we have to an AGI, just finds which word is statistically the most appropriate given what came before and what it's responding to. It doesn't actually know what it's saying rather it's just finding out what humans have said about it and regurgitate it, likely in a more optimal format.
It'd only become worse when Humans are reduced to niche uses wherein most of the content is ai made leaving little in the way of 'real' inputs to use possibly resulting in a slow negative feedback loop of it's quality.
Filtering would likely merely enrich what it is working with but it would hardly help it come up with something new, creative or fundamentally better than what's come before. It will remain creatively stagnant nonetheless. Just cheap, fast and satisfactory.
You forget about fast idiots who call themselves professionals just because they could read how to use the text generator.
Very true
@@OgdenM And even after you train it, it messes up. Again, and again and again.
I think for the foreseeable future, ai is still going to require moderators.
Simple things that we wouldn't overlook would be potentially catastrophic if easily overlooked with AI. For example. There was a higher than human detection of cancer cells of the test data in a particular test, BUT, wound up wasn't looking for cancer cells, it was looking for what machine the image was created on, as the cancerous test data was known ones, from certain machines, mixed in with unknown from newer machines.
It's a game of 'one of these things does not belong' but in that case, that one thing coincidentally wasn't what they were actually looking for.
There's going to be a lot more of that, for a long time, still.
I can give you an exemple for developers. AIs are not yet perfect to write code BUT a senior developer can use AI to write the code faster than any human can, and then debug the code in a few minutes. The code is not yet perfect, but it's close enough to be usable.
Companies will no longer need an army of developers to do the job. AIs will do the heavy lifting and a few senior developers will debug and adapt the code to be functionnal.
Looking at how things are progressing and a constant accelerating pace right now with everything from AI to the rumored room-temperature superconducting material, we're really at a place where we need to ask a lot of very big questions that we've only asked in science fiction until now. You'd think the fact that we've been thinking about these things for a few decades in the arts and fiction would leave us much better prepared but that hasn't really shown itself to be the case.
On God.
People just ignored alo of the last science fiction of the past 80 years in the name of progress.
Late stage capitalism is really a plague
@chandler2 even with a lot of sci-fi claims, some of them probably aren't fake something is gonna happen in the coming years that will change a lot of shit
Oh we are prepared. Its just that the people with money don't care.
We've been thinking about them forever, but we didn't know enough. The world we imagined wasn't as crazy as the world we're in now.
One thing that comes to mind is that although AI is so much faster and usually more accurate, it can often lack common sense and thus jobs would not disappear, but would be changed to supervising the AIs to make sure they don't make ridiculous mistakes.
You wouldnt need as many people supervising the AIs as you would need to do those jobs without AI, so the total number of jobs would still decline
That will only become less of an issue over time, and I don't think it's going to be as long as some people think before the supervisors are the ones needing supervision to prevent them screwing up.
common sense is a dumb phrase
Common sense is sadly very uncommon, and AI is about as intelligent as a rock atm, no matter what anyone says. Ai is learning patterns, yes, but it cant think, it cant create. It doesnt compare to humans. People trying to replace humans with Ai ANYWHERE are gonna hit a wall with it
Also AI art is fucking stupid and shameless thievery of others skill
@@TowerArcanaCrow There's a good interview with mo gawdat (Ex Google x ceo or smth I can't remember for sure) talking about AI that's about an hour long (And he wrote a book about it as well). The situation seems to be a lot scarier than your "about as smart as a rock" statement makes it seem I'm afraid
And if we don't solve the hallucination problem - which we don't know how to, because there is no algorithm for truth - it will unleash an epidemic of truth decay like we've never seen.
I can see a change coming, but the idea of AI becoming more viable because the term was slapped on a language model, though alluring, simply can’t mean that change is happening too soon. Nothing about ChatGPT or any of the similar models are AI.
I’m not in the field, so I don’t know what’s being worked on in the actual field of AI, so it’s likely better models of ChatGPT might end up bridging our massive gap to AI, but I think you question and reason for the video still stands. This does seem like a realistic future and it would be better if society prepared instead of reacted.
Thankfully, this is still very much in “we’ve achieved usable fusion” timetables.
We've all been born around one of the greatest turning points in human history. So much technology advancing so quickly, and more quickly as time goes on. We're lucky to be here. Regardless of how it all shakes out - self-destruction, or self-actualization - I'm glad to be a witness.
Or more likely after thousands of years humans learnt to cure all disease and live forever. Their favourite pastime is to enter a simulation called “Earths golden age” and experience what it was like to live without AI and have a job, go to shops,school etc. Feels pretty real doesn’t it.
Me too,but we are surely walking to the grave.
@@punhosdepano5805 We always have been
its both a magnificantly beautiful and an absolutely terrifying thing, and im absolutely here for it.
The boom of AI was crazy, but now... It feels like its slowing down a lot and what does come out is super clunky and just flat out doesn't work
The only jobs left with be the high ones, the ones at the top, because even with AI someone has to decide if what is created is moral, so Judges, Animation Supervisors, Movie Directors, Chief of Police, all of those will havr to stay, but every entry level, low paying job will slowly dissapear as it can be done better, faster and with the same quality as AI. A massive revolution is upon us.
Love that "Thanks for watching". The best!
There's a recent essay. "The Curse of Recursion: Training on Generated Data Makes Models Forget" that I would heavily recommend reading. I believe it came out of Cambridge. It offers some insight that may at least be comforting in light of all this. It's also a fantastic read that I'd heavily recommend.
Thanks
I would also like to point out that there are a number of recent experimental studies that use synthetic data heavily and sill make improvements, however.
They are actually having a lot of success with using AI output to train AI.
Generative AI is literally going to kill us all with infinite AI-generated MLP fan-fiction.
Might even wind up with a CelestAI.
The bronies really did take over after all lmao
Well, I'd rather live in world like that than one where the ultra rich stay that way off the exploited poor.
2010 called, they want their low-hanging fruit back
Now we need an AI generated jar...
The reason companies won't axe their entire workforce for bots is the same reason Henry Ford was on the side of unions in giving workers Saturdays off. If no one is employed then no one has money to buy the product. If no one has time outside of work, no one will buy the product.
So either we get UBI, or Bots don't take over the workforce.
Actually I already advertise my work as AI-free. I'm a childrens book illustrator and immediately had a flood of people trying to use midjourney and the rest to emulate the style but at least so far its hardly indistinguishable.
I would love see how an AI would behavior manage a rambunctious class of grade 7s while providing social supports for the students. Teaching isn't just about learning and processing facts and data (for humans at least).
Deadly neurotoxins mostly.
Education: last human bastion?
That would be assuming if AI got so efficient, humans would still have a demand for education.
AI doesn't have to teach entire class all at once, it could teach every kid individually.
But the mandate is to get every child in a cohort to a certain level of competence every year. While US schools aren’t doing that very well how would we train an AI with the right data to do it better? Keep in mind that we’d need something with completely different goals than chatgpt.
Having just graduated high school. It's hard to plan for the future like many generations have in the past when issues like these are an everyday occurrence. The question we have to ask ourselves is not, what is going on, but more what can be done. These systems are so integrated already into so many lives that it is difficult to really get a grasp of the magnitude. Like mycelium in a forest where you only see a few mushrooms in the dirt but underneath is a massive sprawling ecosystem, already thriving. And as someone who is going into a creative field, but also with friends going into STEM fields, it's concerning, intriguing, exciting, and everything in-between.
Just remember that life is about the relationships u have with the ones u love, therein is where u find meaning.
"The question we have to ask ourselves is not, what is going on, but more what can be done." You're already on the right because that's exactly the kind of questions and thinking that will guide you. For the past century and longer, we have all been faced with similar challenges and questions.
Some people might even ask themselves, but never actually try to seek out answers. What do I want to do? Where can I have an impact that's fulfilling? Where do I want to go and who do I want to meet? Seek out those answers for yourself and you'll be OK.
Hot tip though - I graduated nearly 20 years ago and had the exact same “how the heck do I plan for the future”. No such thing as social networks, the internet was slow and still essentially text based.
My best advice is “just keep swimming.” 😂 no one really knows where any of this tech will go but you sound like you’ve got a great attitude!
You have planetary extinction to consider.
Not that humans see far enough into the future.
You all died then... Today is just a simulation of the past.
@@Matt-ym3if
Not really... Mankind is dead to me and I have no real connection to this world.
I hope this iteration survives or you all will be erased like a dream.
FanTASTIC vid. Absolutely wonderful, thought-provoking, contemplative, etc.
Damn, and I thought I was being wryly amusing by thanking Alexa and hearing her answers (which are weighted, BTW).
The scary thing is that we soon WON'T need each other, thanks to an uberly-programmed AI.
It still needs the human touch to synthesize and integrate information.
we still need each other to try to push these innovations in the right direction, as well as to stay together as to keep the world from becoming a worse place than its already become.
As a truck driver I always shake my head at the idea that " a robot will never drive a truck"
"What do such machines really do? They increase the number of things we can do without thinking. Things we do without thinking, there’s the real danger." Frank Herbert
Not true just a quote bro
A calculator doesn’t make me lazier it makes me smarter
Calculator is just an example for a bigger concept. With my calculator I am able to calculate things with 2x the accuracy and time as someone who doesn’t have one
@@ALONSO.7.7 Your example doesn't work here, "machines" are pointing to things that "help" us do things without thinking, which to me is as in making decisions. A calculator doesn't make you smarter, it frees you from the work you should've taken hours and focuses you on other stuff. But here's the problem, it made you stop thinking about the intricate processes. You stopped thinking about "How does this equation work" or "How did they come up/prove this equation". The hint of similarity to the quote is there, but the significance is not.
Though, it in fact did improve your speed and productivity, like every widely used tool there is. This is why I don't see companies or even normal users abandoning it soon.
@@ALONSO.7.7
it just makes you aware of the answer!!!
Smart
is the working out part of the answer !!!!
People that discuss this topic are often those who know computers and how to use them well. Even with incredible learning abilities and algorithms, a computer/programs is only a smart as it's user. When I say this, it mostly refers to things such as doctors and creative ventures. There's also the possibility of feedback loops that create false "positives", especially in the creative field. Heck I've seen this with animals as humans made by AI videos. The male portraits tend to be more varied while the ones of women tend to be very genetically pretty. These portraits are then put into the algorithm and, in turn, skew the sampling size for what's generated. Humans are prone to this, by all means. It's literally what in fashion styles, trends, and even memes. As for AI making "a more beautiful x than a human ever could", I'm iffy on that. Is it possible? Sure. But taste is also subjective to the consumer.
My actual biggest concern is the potential for AI to be used and abused for ill gotten gains, to spread false information (especially the type that starts with a grain of truth and falsely draws conclusions from said grain), and intrusive surveillance.
I worked in customer service and tried to help set up AI. At minimum, we wanted it to help with the busywork like ticketing and basic frontline customer service and software troubleshooting; at best, we wanted it replace us. It couldn't even handle basic ticketing. I think eventually it will get there but it will be a while yet. It will take software engineers to guide and create AI.
I'm hoping this AI revolution will work like the invention of the lightbulb and the industrial revolution and similar periods of time. Make us more productive so we can branch into other things. Like previously said, it can possibly make working for a living moot and society can turn in another direction. On the other hand, in today's society there are those that will work as hard as they can to use AI and make themselves wealthy while keeping everyone else struggling. I just hope the right people come out on top.
best we can do as a people is steer it in the right direction and educate people how to treat the increasing uncertainty.
9:44 A section of my family, even the one's working. Don't have access to most online purchases.
As a credit card isn't essential like the States.
The day these ai were made public was the day we forever lost the ability to tell if something was made by a real person or not
Yeah its going to change social media completely, why bother having a conversation with something that you don't know is human. People usually tire of arguing after a half a dozen exchanges, but bots will never tire, and will push nefarious agendas to persuade the public at large.
That's one of my concerns. Another one is AI fabricating an "event" that never happened yet a political reaction happens due to it; where anyone who denies it happening are labelled as spreading "misinformation" or something. We are always at risk of living in the cave and watching the projections cast on the wall, yet this may make it worse.
@@theterminaldave bots have been doing stuff like this for a long time, it's always been pretty cheap to get youtube bots to upvote your videos or spam propaganda in comment sections
Metal Gear Solid 2 outright predicted this future.
@@epsilon752along with bots photoshop has been around for a while. People have been fooling people with photoshop generated images too
The issue I see right now is there are fundamental flaws in GPT and other GPT-like programs, the hallucinations. They can't be removed since they're a part of how the program works. It's not grounded to our world, it guesses at our world and makes mistakes because of that. Nothing I've seen has a solution to that.
Not to mention that since ChatGPT has been released and used it's own outputs are getting mixed together with training data (since they just scrape everything they can off the internet) causing it to start to actually get worse
It's not unique to the large language models. Human brains do that too - with varying levels of self-awareness. Just asking existing models to self-evaluate their processes already makes the output better. Building that into the models more thoroughly will surely help further. But you're right, it's always going to be there - it might just be a flaw in any such system, biological or electronic.
People do this too though..
It’s like self driving cars, your argument is like saying “but there’s a fundamental problem, they can still crash”.
What if LLM’s hallucinate less than humans?
People replying like "humans hallucinate, too". First off, AI hallucinations aren't like human hallucinations. "Hallucinations" are what we call it when AIs start making shit up, address an "easier" prompt, or start rattling off random stuff unrelated to the prompt.
Second off, the reason why these hallucinations are endemic to current LLMs is because there's nothing to ever tether them to reality. They make EVERYTHING up, and only the testing machine (which isn't included in live models) would have any way to check validity.
The difference is that humans have actual intelligence. LLM guess words based on a predetermined system of weights and measures. It can't learn, adapt, or improvise. There's no intellect there, just a black box machine that no one really understands. And if anyone fires back "but we don't understand human intelligence either", that isn't an argument that supports anything. If anything, it's just further evidence that we're putting the cart WAY before the horse.
@@EvilAng3la The difference is that a human can tell when they aren't sure of something. If I'm asked a question about the history of Estonia, I wouldn't just confidently give an answer that I had made up off the top of my head, or "hallucinated," as it were. I would say "I don't know," and if it were important that I provide an answer, I would go do research into that specific topic. Problem is that the AI doesn't know if what it says is true or not, all it knows is that the words it's spitting out have a high probability of appearing next to each other based on their weighted values in relation to the input given. When a model is trained, it's being trained on words, not on the meaning of those words. It essentially assumes all words are equally true. To say "oh humans do this too," is ridiculous. A human can judge whether something is likely true or not based on past experiences and knowledge, and if they aren't totally positive, they can look it up and check. You could say, "but people say or believe things that are untrue all the time without looking into it," and sure, but the key thing is the human has the capability of doing so, whereas an LLM does not. An LLM does not have experience or knowledge, it is a mathematical algorithm assigning weights to words based on their frequency and relationships. Did you people even watch Kyle's video explaining what it is? Because it seems like you didn't.
I love your videos, and i believe in this message and watching the immense amount of capital being jammed into research and design both on the software and the daunting levels of expanding hardware. So much so i wish i could share this clip to everyone i know like at work and such. Ive been too embarrassed todo so though with the midroll ad break about the ball shaving sponsor. :S
Thank you for your continued education of the masses.