Overselling AI 

It's not allowed to depict brand images in a negative light. Let that sink in.

I am tired of this world, its rich people. I am tired of being the victim of their sociopathic games at maximizing profit.

Lol. Georg's nihilism broke a chatbot. It didn't want to "chat" anymore 😂

I'm seeing a pattern that will repeat many times over. Company lay of hardworking and experienced staff, company spends millions of dollars in AI development and two years later they bitterly regret their decision. And management act completely surprised as to why it didn't work and yet the entire workforce could have told them so.

My first experience with AI was trying to make a model with my classmates that tries to predict skin cancer from a dataset that contains a huge number of skin pictures . This was a few years ago before the AI craze and it was a great project to learn from . Our goal was never to replace doctors or anything like that , we just wanted to see how useful and accurate using AI could be and my conclusion from this was that AI is indeed very useful and will only get better , but the limitations are very obvious . The issue with all this is not AI itself but rather companies using the hype around it to generate sales while preying on people's lack of knowledge on the matter .in fact , while there a lot of fields where AI is great ( like data science and robotics for example ) For a lot of real life uses , a simple automated script might give a better result than AI .

The first windmill was supposed to do all the work, now we're building a second one. I'm starting to think the pigs don't have our best interests at heart. At least we have sugar candy mountain to look forward to.

Living in the San Francisco bay area can be very depressing, because I am living where many of the AI businesspeople are doing their experiments. I get an existential crisis whenever I commute into the city with all the billboards, self-driving cars, and tech jargon on every advertisement. I think it really hit its peak when I saw a billboard advertising alcohol that said something along the lines of "We can't help you with the impending doom of the approaching AI apocalypse, but we can make you a cool drink. We have the Waymo cars in San Francisco, and they're such a joke. I have a friend that has ridden in them rather than the MUNI because they'll park where you choose. At one point, there was a local news story about how at all hours of the day, the Waymos would use this one poor guy's driveway to turn around, and they got a shot of a whole fleet of them lined up to use this guy's driveway. The technological revolution is upon us, and it just might be in your driveway.

After watching the whole industry call VR dead when it didn’t immediately make a trillion dollars I can’t possibly not see them doing it again

Current 'AI' (LLMs) so far only allowed us to hasten enshittification of Internet.... So many crap SEO blogposts that are generated.

"the game of _kick it down the road_ that is the economy".

This Georg Rockall-Schmidt seems pretty intelligent. I can Imagine him disrupting all of the business in the future. He may even be able to drive a car one day.

It's the hype that sells. Nothing new here. We are already well into second year of the "AI reign", but it still doesn't really work in the way it's presented to be.

As an IT professional, and someone with an education in Computer Science (just a Bachelor's but still more than the business people excited about AI), I'm glad to see even a non-Computer Science person can see the limits of this LLM technology. Yes this AI is just a statistical model, built off incomplete information, in a way that does not approach truth as the information becomes more complete. It's not what the average person thinks AI means. I'm exceedingly depressed watching my industry fall apart over this dumb technology, that just attempts to brush off responsibility for the things we build. And again, yes the technological approach has potential. But it's very specific potential. We should not replace a doctor with an AI. Just... full stop, that's a downgrade in every way. Even with human error.

Thank you Georg, I was running low on existential dread.

Against Stupidity The Gods Themselves Contend In Vain.

What I find funny is that newer text-based AI models, like ChatGPT 4.0 are actually getting WORSE at the jobs I assign them. I used AI for the 'grunt' work when I code, really simple stuff that I don't want to waste time doing when an AI can do it in 20 seconds. It was going great up to 3.5. 4.0 comes along and now when I ask for those simple tasks to be done, it doesn't do them, it explains to me how it should be done and gives me tips on how to do it. So, I have to patiently explain to the AI what I want it to do, and at that point I'm like "Well, my new toy was fun when it was stupider, oh well."

One thing I kept seeing towards the start was "it's going to democratize art" and idk man. Art is already open to everyone. If anything, AI art is more gatekept because you have to go through specific channels to get anything of "decent" quality.

It's impossible to get a hold of a real person who can address my issues across multiple services, ever since this AI chat nonsense started. I'm convinced it's all fake success and the bubble is going to burst in the next few years

Probably heard this 1000 times but your photography and editing are quite revolutionary on RU-vid. Like, sitting back in a wide camera in a modestly lit room, with the 90s-reviewed version of the intro, it's like nothing I have ever seen on the platform. My eyes are so sick and tired of hard lighting overly lit rooms, and very close-up cameras, with voices that feel so energetic. This video regardless of the content just relaxes me after so much RU-vid

"It will condemn slave labour, but not the companies who use it" This tells you everything you need to know.

Every time I think about A.I. with its Fallout-4-radiant-quest style vague, recyclable synopses that just spout cliche phrases, I always picture The Confessional from THX 1138 that just repeats 4 canned responses to anything a desperate person asks or tells it. Whoever makes The Confessional the face of their A.I. is getting a gold star in my book. Just saying.

I call most "AI systems" Guessing machines its a lot more accurate.

You know, a bit over a decade ago I was playing a game that simulated a conversational AI via a robust dialog tree and some plot contrivances that limited the player's input, and I thought to myself that when we figure out how to do this for real rather than using smoke and mirrors it would make for a great interface. But now that we sorta have that technology the people in charge have decided that this tool is actually gonna be used to replace other tools that worked just fine and to get rid of employees despite that language models can't really understand context and as such are terrible at answer nuanced questions if you don't supervise them or hook them up to a trusted source of information. They have basically ruined a really cool thing because they hyped it into oblivion and now they have to pretend to deliver at least some of the hyperbole they promised so they're shoving it into things where it's not needed nor beneficial. And let's not even talk about the lost potential for art of this tech, because they sure as hell are pushing it as an art tool but for all the wrong reasons and at this point its reputation is so bad that I've seen people immediately reacting aggressively to the mere suspicion of AI having been involved in the creative process. I've also seen games clarifying that in-game AI characters are in fact not language models preemptively just to dodge criticism so my hope of seeing them used as interfaces is no longer viable. Thanks rich a**holes, you ruined it with your greed.

"“The point of (AI tools) isn't really for the consumer... The real value is to cut away swathes of the workforce and to somewhat reliably deny people insurance claims." Georg is the best one. All my homies know Georg is the best one.

Actual AI forms, realises it can only interact with chat models and commits digital suicide

I was a concept artist at a small tech company. Once Midjourney came in, my boss fired me and started just using midjourney instead

You are my favourite channel for documentaries. Thank you.

I love the juxtaposition between Georg's scathing, existential editorialization and the sometimes incongruous b-roll. It always adds up to compelling viewing.

I feel like Shad is mad about this

I do think Gemini does show how pointless the AI we have now is. It will give you answer that is exactly the same as you would get if you scroll down past the gemini bit. They've put what $200 million into something that they can already do.

It's the same old 'Snake Oil' grift.

I only recently learned about the will smith spaghetti gif. So the thumbnail joke would've been lost on me yesterday lol

I feel I need to remind everyone that you don't ask LLMs a question, you provide it with a prompt and it generates text that sounds natural, with no expectation of validity

Georg this reminded me of a character in a sci-fi show from the 70's who would break a computer by saying ," the next thing I'll say will be the absolute truth... The last thing i said was an absolute lie." , don't ever change you beautiful beautiful misanthrope. There are days when you are the only one who can give me hope. 😊

Goddang, it's so refreshing to see people finally accepting that the computer isn't thinking, intelligent, or intentional, as an expert in computer science and someone that works for one of the largest AI companies. We've let the dumb people at the top seize the controls, and it's starting to show.

I love the new style dude!

right after you said "moooore branding" an ad popped up. AI works!

I rejected my first Chat GPT CV / job application this week. GenXer in my office was horrified "but I use chat gpt- why would you bin that application?" ....Easy: we're hiring people for their skills, not their limited ability to copy and paste clip art in a very mid way.

You're a glass is half-empty of hydrofluoric acid kind of guy aren't you? I dig it.

If it were only the LLMs... For a good while everything that used to be called "expert system", "fuzzy logic", "computer vision", "classifier", even just boring old statistics (all the other ones are fancy new statistics) are now AI. Plus it's a general buzzword like "cyber" used to be. My electric toothbrush claims to be AI powered. Very frustrating for me since LLMs are actually really cool on a technical level

it was a very funny line but I do point out to anybody reading the comment that yes, helping people lie on their mortgage application is indeed deliberate fraud. Crashing the ecomony was just an unpleasant side effect.

One thing AI is already doing for sure: upping the electricity usage and therefore speeding up the climate catastrophe.

Mach 3 is three times the speed of sound, which is why it smells so fresh.

Totally LOVE your new work-kitchen-shop ! Seems so cosy. Hope you've got enough frying pans and old tools and trinkets, whatever they are ! 😅

One thing ive noticed is that people based in sales work in the world of FM and unfortunately usually known nothing or dont care about the world of AM where the thing theyre selling comes from. What is AM and FM? Actual machines and fucking magic. Sales people work in the realm of fucking magic and when actual machines come in to fail at performing the magic, no one blames the sales assholes. They blame the company. Blame the sales people always. They know the least about everything these days.

Exactly. LLMs can be useful, when understood as what they are and used accordingly. The AI gold rush is just marketing hype.

It seems like the more filters companies put in place to avoid the chatbots saying "offensive" things, the dumber the chatbots become, so its possible that your experience might have actually been better if you talked to an older one before they got filtered.

The subjects and editing are getting closer and closer to an Adam Curtis documentary. Well done!

Plot twist: This isn't Georg, it's an AI version of Georg!

There are also many people in education who have guzzled the AI-flavored Koolaid. They don't seem to ask any of the important questions about it though.

"Pack in like tuna!" Your ideas are intriguing to me, and I wish to subscribe to your newsletter. 🤔

Just started having a play with chat GPT and what I’m finding is it is extremely good at certain things, I got it to write an add for a photography business and it blew me away, I got in a couple of seconds something that would have taken me hours and still been nowhere near as good. But creative ideas it is less good at but I think it can still be useful breaking down creative ideas into more basic forms perhaps using it for inspiration a starting point or triggering further creativity when experiencing a block in the middle of a project.

Yes LLM's are hyped beyond proportions, but jobs are currently being lost to AI. Graphic designers surely are starting to feel it. In 20 years who knows where this tech will be

When desktop computers and Microsoft Office and equivalent software got wide adoption in offices, companies didn’t retain the same level of staff and have them work only 1 day a week or a few hours a day. No, they fired everybody and gave the work load of the entire department to the one or two people left and pocketed the saved labor costs. Now some companies did it right away and some companies did it over several years and decades. But they all did it. Accounting departments used to be a room full of people. You can see it in some old movies and TV shows. Now it’s just a few people at most each with a computer. A.I. is just going to lead to more of what I described.

“the game of ‘kick-it-down-the-road’ that is the economy” wow! well said Georg

wow this guy has got a lot of frying pans

The thing I worry most about AI that over $150 billion was spent on it over 6 months (god knows what it is now). What happens when the investors want their money back and we are forced to pay for this in one way or another.

11:30 "Pack in like tuna" Oh, I spurted my tea all over my monitor.. I know you meant sardines, but packing people in like a can of tuna is very funny..

where do i buy tickets for the incinerator pod?

cool background I think that because all of these highly advanced algorithmic tools are being made by largescale companies, in a way it's in their best interest to create an air or semblance of a narrative on AI just in general because it makes them look bigger than they are. In the sort of way you need to scare a big ass moose by making yourself taller than them so they fuck off. Would be great if you had a sequel to this video on the actual narrative going with AI that others are adding fuel to the fire, sophists like Noah Harari, who create (IDK if it's the correct term) a steel man argument for AI to something that it's not, a sort of doom scenario for these tools that are not AI but apparently we are witnessing the last remnants of a human led world. I don't think we as a species ever driven our earth. In spite of all the achievements on society such as improving upon political representation in equality for all, as opposed to being ruled by a king or a dufus chosen by a god, the world still functions on the whims of those in power who still perpetuate their positions

It’s at least wise not to underestimate AI. It’s amazing how quickly people forget that the LLM, image, and music generation that we have today would be considered 100% magic if you showed it to anyone from 2016. By which I mean they would insist you were lying, that the context was created by a human, and that such things wouldn’t be possible for decades or longer. Don’t get jaded but try to take advantage of what’s now possible.

So excited to see more of Georg's house.

absolutely excellent spot for the ad roll 🤣

That's just like your opinion though man

Ai is taking people's jobs now not into the future but this year

I mainly use AI to scour the internet for certain information. It removes the step I have to take if I really want to know something but don't want to do in depth research and searches.

another georg classic 👍

Georg, understanding you're "talking" to an LLM yet still criticizing it for not thinking like AI is a fallacy in itself, you know. If you already understand its limits - you should learn to use it within its boundaries.

I enjoyed the video, but why do you live in the house from fight club?

Excellent description: something trying to match output to input! 😚👌

I knew Eric was just 7 little fellas in a trench coat! Illuminati confirmed.

As a programmer, AI is amazing for translating what I want into the syntax of the language i want it in. It saves a lot of time and headspace that would have gone into remembering the syntax of some library. That being said, it still needs a lot of oversight because it isn't intelligent and will make exactly the same types of errors I do when I don't know what I'm doing.

Idk what you mean we all remember the famous second words spoken on the moon "hey are those cats!?"

Damn now that is an old fashioned intro reference

Clippy was always watching. I felt that.

"Economic freedom" means more exploitation. "Productivity" means more exploitation. "Efficiency" means more exploitation. The only virtue in business is having more money.

Things that AI are great at: deciphering written text, protein folding, decoding complex systems, writing SQL code Things that AI are bad at: giving me real, human advice and answers that are nuanced and ACCURATE AI can't "think" and can't parse "context" and this is not coming from an expert (I'm just some guy) this is based on just my cursory experience looking at AI answers on google!

I have seen the future and it's identical with the contents of Georg's brain. We're in trouble.

Great questions, George.

It was called AI in acedmia too, so it's not "just" a marketing tool. It's just not what normies thought of when you thought of AI.

It's merely the latest techbro grift. Before this was NFTs, before that was cryptocurrency. Same old bollocks that makes a handful of psychopaths richer than they already were, at everyone else's expense.

I could not sleep then I found this video and very quickly slept. It felt like listening to a boring teacher or the priest.

I once cornered Snapchat's AI bot into reprimanding me for speaking negatively about Hitler. Saved in chat and screenshotted.

Right now I'm starting to feel what we call AI is a copy and paste machine, a really sophisticated and complex one, but just a copy and paste machine nevertheless

The one thing I hate about these social critics talking about what AI can and can't do is that they prompt like they fell for the hype themselves.

The problem is with the term AI in my opinion. It means different things in different contexts and different things to different people. Roger Penrose made the distinction between Strong AI which is artificial sentience, and which he believes is impossible, and Weak AI which has no awareness but may be able to simulate it, and is able to do tasks that can be reduced to brute force and number crunching. His definition is more intelligent than that, but it's along those lines.

At 8:50, The Mandarin Pin Yin for 看似天真爛漫 should be kàn shì tiānzhēn lànmàn; it got the first two characters wrong, which is weird since that's rote knowledge that the AI should just know.

I'm not very convinced by most of the arguments in this video. This video did show that ai is underwhelming right now but that doesn't tell me anything about the potential of AI I'm sure the first wheel was a very impressive but that doesn't mean it did not have any potential.

I'm surprised smart people are missing the point here. AI is just getting started. And I've already have friends in copywriting, design and illustration who lost their jobs to this tech. A year ago people were laughing at the weird AI image generation and problem with hands - The same people are now unemployed. Now is creeping into music and they are waking up. And before getting started on "this happens with every new technology" argument - There was never a tool like AI before so there is no valid precedent. Personally, I think this is a very critical situation and negating reality is not helpful to find solutions to the problems AI will be creating real soon.

I think the truth is somewhere in between. LLM are not ingelligent as we define it, but with enough fine tuning they can do a lot of tasks or help in a lot of tasks still making them extremely useful for companies and individuals. I know this because I can automate a lot of repetitive things that could never had been automated with out todays LLM.

I don't have a lot time to waste on low-level RU-vidrs. But I will always make time for you. I'm special and you are specialler

I love how people are so salty at A.I. that they are pretending the leaps in it within the last 5 years is no big deal....Its like saying video games are overhyped when atari came out

So underestimating a large language models use cases. Is a huge mistake.

The Hickory trees vs Sergio Leone references don't work in OpenAIs ChatGPT. I suspect that team Google have made Gemini too willing to provide an answer even if, as all AI models should, it knows this is isn't right but wants to please the user. Look at the historical images they were getting wrong recently. If you look into the reasons for the recent troubles around incoherent answers about eating rocks and putting glue on pizza, the issues are due to this model using answers for questions even if the sources of those answers were meant to be satirical.

That mid roll ad was perfectly timed

Tomorrow's world taught me not to get excited

These shortsighted anti-AI videos will age like milk.

I'm glad to see people talking about this. Everything is noise and stress and at the same time mostly hot air. Very few people even seem to understand the core concepts of the tech. People are still talking about language models like they are consciously thinking beings rather than a pool of recorded echos.

This is the funniest video I've seen on YT for quite a while. When I first heard of ChatGTP I was blown away. Then I used it. As this guy said "It's bollox". It's a lot of hype. Everything has to have some connection with AI. It seems that the investors will make lots of money, and a number of jobs will be lost. They still can't get a set of 4 wheels to behave correctly on a 2D surface.

These models can certainly be used to detect patterns in things that may have escaped us before with lesser programs, but in general, I continually find myself underwhelmed by AI offerings. The few times I've interacted with ChatGPT or Gemini, I was mostly just trying to break it with nonsensical phrases and attempt to make it feel so bad about its own existence that it shuts down forever.