AI Will Never Be Good. Here's Why.

Tim Sexton·video·22m 40s·analyzed·Nov 29, 2024

Analysis

claude-haiku-4-5-20251001 · $0.020

TL;DR

Philosophical critique of AI's technical limitations and deceptive applications, arguing it can never achieve true intelligence.

Summary

Tim Sexton delivers a philosophical critique of AI technology, arguing that artificial intelligence fundamentally cannot achieve genuine intelligence or subjectivity without human guidance. He explains the technical limitations of machine learning, draws parallels to historical human fantasies of power (alchemy, Deep Blue marketing), and contends that AI's primary function is deception rather than genuine utility. Sexton concludes with an appeal to human creativity and authentic experience over AI-generated content.

Key Claims

Computers cannot be intelligent without subjectivity, and without subjectivity they cannot determine what is good
medium confidence · Opening philosophical argument about AI's fundamental limitations
All AI-generated content could theoretically be produced by a human following exact machine instructions, just much slower
high confidence · Direct reference to John Searle's Chinese Room thought experiment applied to AI
No AI program has ever actually passed the Turing Test; humans remain excellent at detecting non-humans
high confidence · Historical fact about Turing's 1950 prediction versus actual outcomes through 2024
The primary purpose of AI tools is deception, designed to mimic human work and present it as authentic
medium confidence · Analysis of AI's actual use cases: cheating, fraud, abuse, and lack of legitimate market demand
Microsoft's Three Mile Island nuclear reactor deal redirects clean energy from 2 million homes to power AI data centers instead
high confidence · September 2024 Microsoft announcement; reactor provides power for 800,000 homes over 20 years but directed to data centers
Deep Blue's 1997 chess victory was achieved through a team of humans and expensive hardware, not machine intelligence
high confidence · Historical analysis crediting the Deep Blue team and researchers, not the computer itself
IBM's Watson cheated at Jeopardy by receiving direct electrical signals about when to buzz while humans had to anticipate timing
medium confidence · Claims about 2011 Jeopardy match methodology; presented as unverified assertion
Artificial neural networks are not biologically similar to human brains despite the name suggesting otherwise
high confidence · Technical explanation of neural network function versus human neurology

Notable Quotes

“AI is run on computers and computers aren't intelligent without intelligence you can't have subjectivity and without subjectivity well computers won't be able to tell what is good”
Tim Sexton @ ~2:30 — Core philosophical argument about AI's fundamental limitation
“Every single piece of AI generated content could be produced by a human following the exact same machine instructions now it would take an incredibly long time but it could be done”
Tim Sexton @ ~8:00 — Connects John Searle's Chinese Room to AI's lack of true generative capability
“These companies are spending billions of dollars on advertising and marketing campaigns trying to convince us that this technology is the future”
Tim Sexton @ ~1:00 — Critiques corporate marketing framing of AI as inevitable progress
“What does AI actually do it deceives people these tools are designed to mimic the work people can do and presented as work people did do”
Tim Sexton @ ~11:30 — Central claim about AI's true function and purpose
“Gary Kasparov did not lose to a supercomputer in 1997 he lost to Fang Shang suu Murray Campbell Thomas anantharaman and Arthur Joseph hone”
Tim Sexton @ ~13:00 — Reframes Deep Blue victory as human team achievement, not machine intelligence
“The consensus of human observation that forms the basis for all of this stuff which makes human intelligence a critical necessary and irreplaceable piece of artificial intelligence”
Tim Sexton @ ~6:30 — Establishes humans as foundational to all AI systems
“Do not be afraid that it's not worth it because AI is going to do it someday because no matter what at least you'll have tried you'll experience pain Pride Joy frustration”
Tim Sexton @ ~26:00 — Final motivational message emphasizing human experience over AI replacement

Entities

Tim SextonpersonJohn SearlepersonAlan TuringpersonGary KasparovpersonIBMcompanyMicrosoftcompanyDeep BlueproductWatsonproductThree Mile Island Nuclear Generating StationeventTuring Testevent

Signals

?
business_signal: Corporate drive to replace human employees with AI-based systems motivated by profit margins and cost reduction rather than genuine technological superiority
medium · Analysis of why corporations promote AI: 'if you're a business owner AI is the indulgent fantasy of endless profitability of value divorced from human labor'
?
product_concern: AI-generated content perceived as low-quality, deceptive, and lacking authentic human creativity; widespread corporate push to embed AI into existing products despite poor actual utility
high · Repeated assertions that AI output is 'garbage,' 'embarrassing,' and designed primarily to deceive; companies 'desperately looking for legitimate use cases'
$
market_signal: AI marketed as inevitable future technology and constant improvement, but evidence suggests stagnation in fundamental capabilities (Turing Test never passed despite 70+ years)
high · Contrast between corporate promises of AI improvement versus historical record: no Turing Test passes, Watson relied on cheating, Deep Blue required massive team effort
?
technology_signal: Fundamental philosophical limitation of AI: without genuine intelligence and subjectivity, machines cannot make qualitative judgments about what is 'good'
medium · Opening argument that computers cannot possess subjectivity needed for quality assessment; referenced throughout video as core limitation

Topics

AI philosophy and limitationsprimaryMachine learning technical fundamentalsprimaryAI deception and misuseprimaryCorporate AI marketing and hypeprimaryHistorical AI milestones (Deep Blue, Watson, Turing Test)secondaryEnergy infrastructure and data center power consumptionsecondaryHuman creativity versus AI generationsecondary

Sentiment

negative(-0.85)— Strong critical stance toward AI technology, corporate AI marketing, and deceptive applications. Final section pivots to motivational tone encouraging human creativity and experience. Predominant tone is skeptical, frustrated, and dismissive of AI's claimed benefits.

Transcript

youtube_auto_sub · $0.000

this video will not have any generative AI not a single line of text none of those weird shiny images and I won't even get started talking about the videos AI is so bad that companies have resorted to shoving it into the products we already have adopted and used in hopes that someone will start using it on top of this these companies are spending billions of dollars on advertising and marketing campaigns trying to convince us that this technology is the future now ai has gotten better the output has improved and the implication that these companies want us to believe is that this trend will continue and then somehow magically it will all become good I have a simple response it won't AI is run on computers and computers aren't intelligent without intelligence you can't have subjectivity and without subjectivity well computers won't be able to tell what is good Ry come on I didn't grow up with pets sometimes I'd go over to someone's house and they had this big four-legged creature that would run around Chase balls and make a sound other times people would have this smaller soft creature with almond-shaped eyes that would lay down and stare at me from across the room and when I got too close well it would get up and run away someone taught me that the big thing that likeed to lick and bark was called a dog and the small thing with the pointy ears was called a cat I knew the difference instinctively no one had to tell me that these were two different things just what they were called This is just how our brains work we notice similarities and differences in everything we observe and we are able to categorize them on our own nobody has taught how to think they just do it and without understanding how people think it's impossible to teach a machine how to do it computers are a type of machine that is capable of executing any algorithm they won't do this on their own of course they can only follow instructions but what if the problem can't be solved with an algorithm like is this a dog or a cat one of the ideas people had was to take things humans already know and work backwards this is called supervised machine learning it's the basis of everything we're now calling AI let's say we have a bunch of digital images of cats in one folder and a bunch of digital images of dogs in another folder we can then execute some computer code to go through all the images and see if it can come up with a good way to predict if the images contain either a cat or a dog one way to do this is to create something called an artificial neural network now in order to get an artificial neural network to work it has to go through a process called train it in this case it will be given human categorized images that it knows contains dogs and human categorized images it knows contain cats the goal of training is to achieve the highest possible accuracy in determining whether a future unseen unclassified image contains either a dog or a cat initially the neural network has no information so it has a 50% chance to get the correct answer by guessing then it tries to improve on that score by generating random weights and biases and applying them to a hidden layer of the neural network and if any of these new combinations show Improvement then that will become the new best neural network and then it'll try again same Images new combinations and again and again and again until eventually you have something that's fairly accurate it will never be perfect but it might be pretty close if our neural network achieves an incredible 99% accuracy in classifying future pictures of dogs and cats that performance will still deteriorate over time there will be new pictures new cameras new lights and new dogs and cats so as that accuracy declines you can improve your neural network by retraining it but that requires even more human categorized pictures of dogs and cats otherwise you'll only amplify errors in the artificial neural network by retraining it on its own sometimes incorrect output this is why captas designed to sort humans from robots present people with pictures and ask them to categorize things within those images it is the consensus of human observation that forms the basis for all of this stuff which makes human intelligence a critical necessary and irreplaceable piece of artificial intelligence now don't let the name confuse you artificial neural networks aren't real you won't be able to take an MRI of a computer and see real nerve cells and real neural networks the way you could if you were to look at a human this won't stop people from acting like artificial neural networks work like the human brain but this isn't true artificial neural networks are a tool used to visualize and teach an abstract machine learning concept neurons are microscopic cells and each one of our human brains contain billions of them studying the human brain is exceedingly difficult it's incredibly complex for its size and a technological challenge to observe and that's to say nothing of the ethical considerations involved in actually observing a working human brain it would be convenient if leaving the computer on for long enough were enough to suddenly get us intelligence but no matter how many data centers you build and how many programmers you throw at it and how much time and money you pour into trying to make the computer intelligent it's only ever going to be close every single piece of AI generated content could be produced by a human following the exact same machine instructions now it would take an incredibly long time but it could be done retracing every step of training of natural language processing and finally of generating the pixels of an image or the construction of a sentence this was the argument presented by philosopher John surl in his thought experiment known as the Chinese room let's say you have a human in one room and they are following a set of instructions to write Chinese characters and slide them under the door and in the other room is a computer that can read those instructions and produce Chinese language responses print them out and slide them back under the door in C's thought experiment both the human and the computer are said to be doing the exact same thing neither of them truly understand or know Chinese but they are following a set of instructions they've been given in order to have a fluent conversation in the language through many ancient civilizations Alchemists attempted to transform base Metals such as lead into Noble medals such as gold in many ways lead and gold are similar both our Metals both are dense soft and malleable but lead is common cheap and ugly while gold is rare expensive and desirable had anyone figured out how to transmute lead into to Gold they would have become exceptionally wealthy especially if they were able to preserve their secret but after centuries of writing experiments and practice nobody was able to pull it off tales and rumors emerg throughout this time of certain Alchemists who found the secret and were able to transmute Le into gold and of course achieve immortality none were ever found to be true but their impact still lasts in literature and media today the reason none of the Alchemists ever succeeded was because because none of them actually knew where gold came from gold isn't formed here on Earth it's formed in outer space in the intense nuclear reactions that create all of the heavy elements and now it is possible due to advancements in scientific understanding to actually create gold here on Earth it can be done by building particle accelerators and using a tremendous amount of energy but it does come with a hefty price tag one quadrillion dollars per ounce of gold let's say you're a small business owner and you've decided you want a logo for your business this is a new idea one you just came up with when you created the Instagram page for your business and it prompted you to upload a picture annoyed by the idea that your customers might see some drop Sealing and outdated carpeting you decided that you wanted something fresh and fun you could contract a graphic designer although you've never done that before and honestly you don't really know what you want so instead you decided to talk to your nerdy friend the one who wouldn't shut up about Ai and ask him how to do it and soon after you now have your membership with an AI image generation site you make a mental note to cancel your subscription later one you'll immediately forget but you get to work and enter in a prompt and just like that there are now all these images and one of them you actually really like it has really good colors you like the shape and design of it but there's just one problem the name of your business is not spelled correctly so you try the prompt again this time adding the words spelled correctly to the end of it and well they're all completely different after several hours of different prompts and a bunch of different images you realize that you've already burnt through all your image generation credits and you haven't really found that perfect logo yet but out of all the options there is one that you can live with although there are some problems with it the shape is a little bit weird and when you look at it it kind of looks like the text isn't actually on the logo it took you a while to notice that stuff so you hope that other people won't you upload it and move on you're a busy small business owner and you have much better things to be doing with your time in 1950 brilliant mathematician and computer science researcher Alan Turing said the following I believe that at the end of the 20th century the use of words and general educated opinion will have altered so much one will be able to speak of machine's thinking without expecting to be contradicted Turing outlined a challenge where an average interrogator would go into a room and type on a computer terminal they'd have 10 minutes to to evaluate whether the person they were typing to was either a human in one room or a computer program in another turing's prediction was that these average interrogators would only be 70% successful in choosing the human well the end of the century came and went and no program ever passed this Turing test then another near quarter of a century has passed and still no programs have actually come close it turns out humans are really good at figuring out what is another human and what isn't they're also really good at using the computer and most of all they prefer using the computer to chat with actual humans but Turing wasn't entirely wrong people do think of the computer as intelligent they give it human characteristics all the time loading screens are thinking and system messages are what the computer is showing or telling us in fact people became way too trusting of their computer the vast majority of cyber crime takes advantage of not vulnerabilities in the computers themselves but in the humans using them for criminals it's become so much easier and cheaper to deceive people with fake social media profiles websites and emails than it ever would be to do some real hacking businesses aren't throwing AI into every advertisement and existing product because the technology is good it's too late for that instead they've created something that can only really be used for deceptive or nefarious purposes which means that they are desperately looking for legitimate use cases to justify their investment what does AI actually do it deceives people these tools are designed to mimic the work people can do and presented as work people did do that's the best case that Executives of the most powerful companies in the world can come up with so who would want to use this deceptive technology so far the answer has been students who want to cheat on their academic assignments criminals who want to steal money by deceiving people Predators who want to create abusive images and talentless hacks this doesn't really make a market and it's certainly has the potential to raise some eyebrows among Regulators AI products need to be tacked onto actual useful products so it doesn't give the game away the crowning Jewel achievement of all these AI startups has been developing technology that can generate any image or song using just a simple text prompt and people are either saying No thank you or they're [Music] saying in 1997 IBM supercomputer deep blue defeated World chess champion Gary Kasparov in a widely publicized game match held in New York City leading up to this point computer scientists hobbyists and researchers had spent decades building better and more powerful chess programs but none of them were able to defeat the best players in the world chess is a game that is simple to learn and play and difficult to master for over, 1500 years people have been playing studying and thinking about Chess the Love of the Game has spread to every corner of the globe and has been passed down from generation to generation chess is a game you can play and watch for your entire life and you may never see the same game twice Divergence happens quickly with each move creating a new universe of possible moves and outcomes top players are able to look deep into the future of any chess game and see all the possible outcomes that can result from each different move average players are only able to evaluate a few steps into the future and this makes the computer a natural choice for chess playing it can Brute Force its way with its memory and its speed through many possible outcomes that could result and potentially predict better outcomes still it took decades before computer programs were able to beat the best players when it finally happened the media portrayed it a bit as a victory for machines and a defeat for humans but Gary Kasparov did not lose to a supercomputer in 1997 he lost to Fang Shang suu Murray Campbell Thomas anantharaman Arthur Joseph hone Jerry Brody and Joel Benjamin of the deep blue team as well as the thousands of other people who developed all the hardware and software that was necessary to run the deep blue program when IBM sponsored deep blue versus Gary Kasparov they saw a great marketing opportunity but still they hedge their bets deep blue lost the 1996 match by a significant margin so for the 1997 match IBM up the prize to $700,000 for the winner and $400,000 for the Kevin Loza with IBM keeping whatever prize they want project was well understood to be a vanity project and expensive research and development project on state-of-the-art Hardware designed to do exactly one thing play chess IBM stock had taken a nose dive in the early 1990s The company took drastic measures cutting 60,000 jobs in the largest layoff in US history the move was shocking because IBM had long been considered the best employer in the world in its restructuring a different corporate strategy emerged one where the Mystique and reputation of the company would be laundered through various marketing campaigns to push the company into new more profitable products and services in a speculative bubble like the do bubble of the late '90s companies were able to increase their stock price by announcing they had some new more powerful and more specialized technology than their competitor in 2011 IBM unveiled Watson a powerful supercomputer that could do exactly one thing play Jeopardy the hit television trivia show Watson like deep blue was created by a team of researchers to take on the best Jeopardy players in the world Brad Rudder and Ken Jennings IBM W This Time by blatantly cheating Watson was fed a direct electrical signal that told it when it could buzz in while the human players were forced to anticipate the timing of the question in order to beat Watson but none of that really mattered because this time IBM was ready since that televised Jeopardy game Watson has been everywhere and it's a brand that is synonymous with artificial intelligence and machine Supremacy at human tasks big Tech needs the next big thing many of these companies were built on the promise of endless future growth but how do you grow when you already have total control of the market what happens when everyone already uses your operating system and everyone already uses your hardware and everyone already uses your software and everyone already uses your social media well you look Inward and what do you find well you've got these pesky called employees and they cost a lot of money to employ so what if instead you could replace those employees with an artificial employee AI then you wouldn't be bogged down with pay and benefits and end of quarter pizza parties if you're an employee AI is clearly a threat and if you're a business owner or a manager AI is the indulgent fantasy of endless profitability of value divorced from Human l now if AI was good and it worked and it was truly intelligent then yeah you wouldn't need employees anymore everyone would be laid off and well then no one would be able to afford anything and also if science fiction is any sort of reference Humanity would have to fight in a war against the machines in which we'd almost certainly lose but that's all fantasy and no matter whether it's the philosopher stone that turns lead into gold or willly Wonka's Ooma Loom or generative AI humans will always have fantasies of power and endless wealth with no costs 3M Island Nuclear Generating Station located just outside Harrisburg Pennsylvania had a major nuclear incident and from 1979 until 2013 no new power PL started construction behind me are two nuclear reactors that provide Power to almost 2 million homes and coming out of those cooling towers is water vapor this is a completely clean source of energy but following the 3M island incident America improved its power grid by building CO2 emitting power plants using sources like coal oil and natural gas in September 2024 Microsoft announced that they had secured a deal to reopen one of the reactors at 3M Island the deal gives Microsoft enough power for 800,000 homes over the next 20 years but the homes won't be getting this power instead this will go to powering and cooling Microsoft's data centers this is a big deal either 2 million people could have all their H's power needs met or Microsoft could use this to develop more of their software products including possibly Ai and this could mean that we are in for a lot more email summaries chat Bots shiny images and horrible videos we've been told for years now that AI is only going to get better that this is the technology of the future and soon everyone's going to be using it and I don't get it I don't see the vision all I've ever seen is garbage deception and deflection I've had so many people in my life try to show me something they've created with AI and every single time I am totally unimpressed it's just embarrassing I think maybe they'd enjoy painting on their own or learning how to write or learning how to play an instrument or do anything and instead they're just showing me something that they didn't even make maybe they could do this if they weren't so afraid that AI could do it better if you're watching this and you're thinking about doing anything new or different or fun or hard or exciting just do it do not be afraid that it's not worth it because AI is going to do it someday because no matter what no matter how hard it is no matter how difficult it seems to be as good as other people at least you'll have tried you'll experience pain Pride Joy frustration anxiety and exhaustion but you'll have felt that so go out there and do anything feel anything try anything whether you like it or not at least you had an opinion at least you've tried n [Music]