In a lot of ways it was inevitable America and its Capital region of Israel would become the powers obsessed with Artificial Intelligence…
The Popular conception of artificial intelligence EXACTLY overlaps with the Jewish Golem legend, an artificial black-box slave/robot controlled by Rabbinic Language magic, just so long as you lawyer your wish/prompt exactly right…
Right down to the persistent terror that the Golem will turn on its masters. The Jewish dream/terror that they could or might fail to create the perfect servant for themselves.
But more than this… It is quite inevitable Artificial Intelligence would become an obsession of the American Elite… Given they are the first elite in history to actively criminalize and punish ACTUAL INTELLIGENCE.
Whether it is Grigs vs. Duke Power criminalizing IQ and Aptitude testing in hiring, the defacto mandating of Affirmative Action through the various Civil Rights Acts, or active policies to use IQ testing in government and elsewhere… So as to refuse to hire people who test too highly.
The US has waged a 60+ year war against actual intelligence throughout all its institutions (Gee I wonder why they don’t want pattern matchers in positions of power?)
One might call America’s obsession with artificial intelligence ironic… If not for the fact that the one is an immediate consequence of the other.
.
It would be natural to wonder exactly WHY American society is so obsessed with Artificial Intelligence given the wild misallocation and purposeful sabotaging of actual intelligence throughout American society.
Outside the US Military, where, atleast for now… They can still IQ test a teenager at 18 and have them running a Nuclear Reactor under the sea within 3 years… Nowhere else in the US are you allowed to simply IQ test a heterosexual white male teenager (ideally from subcultures that value honesty) and then train them for whatever advanced skill you need when they test high enough.
There are tens of millions of above average American children 105-145 IQ who could be learning vastly faster and getting advanced language, mathematics, machine skills, and various digital arts education if standards were merely enforced and educational outcomes pursued simply to the level they were in their great grandparent’s days before the Civil Rights Act, when high-school Graduation was an actual challenge only 30ish% of people achieved, and actual skills were learnt there.
The simple fact is that “Mass Democracy” has resulted in the Low intelligence average mass voting themselves the crippling dispossession of the High IQ minority that would be their natural leadership.
What the Majority of Median voters want most of all is for the exceptional individuals to lose… And they don’t care how obscenely they have to rig the game to make it happen.
As soon as Universal Male suffrage, Universal Female Suffrage, and Universal Black suffrage where achieved (and not Universal Immigrant Suffrage… Whether they are legal or not), they immediately began voting in Unison to cripple and prevent high IQ men from working or taking their natural leadership positions in the institutions that quite literally cannot function without them.
Indeed American society is obsessed with sabotaging what few Americans receive an above average education or self-education, and HAVE to be hired by American institutions, by pairing them with low IQ “middle-managers”.
I put “middle-managers” in quotations… Since the majority are not even managers nor do they occupy the middle of a hierarchy, often having no direct subordinates whose results they are responsible for… And shockingly often no direct superior either , or certainly none that pay attention to them.
Instead more often than not they sit as gatekeepers and friction points… Generating meetings and other activity which actually lowers the productivity of others.
Here’s a Fun activity… Next time you’re at a meeting try to add up the cost of the time spent. Guess the salaries of your other employees then divide by 2000 to get the hourly rate (40,000 becomes $20 and hour. 100,000 becomes $50 an hour… really simple trick) then add up the costs of the meeting you just attended. you will be amazed at just how many $1000, $2000 and $5000 meetings you sit through. I once sat through a completely pointless 5 day, 3 hours a day sales seminar talking about completely pointless sales-nothings buzzwords… 30 employees. half making 150k+. The entire seminar series might have cost 30-40k in wasted time alone, and I suspect the cost to hire “the expert” might have been that much or more.
Obviously this was probably a kickback scheme where the Director was converting his “Skills Development” budget into a revenue stream for a local friend who was giving him a kickback or helping him with his next Job-Hop. 100k+ total.
You might react “But that’s an expense the equivalent of 3-5 clients spend!” Ah… But we were a tech company. It was all investor money. The “revenue” didn’t matter for actual cash-flow, it was just to produce a line that would “go up and to the right” so the company’s valuation would go up that we had an “exponential revenue” (all the contracts with clients were structured to produce this illusion, lots of first and second year bargains). The relationship between income and expenses were completely detached from each other… Sure a responsible sole proprietor would look at expenses and treat that the same way as revenue “a penny saved is a penny earned”… But in tech “startup” (it was a decade old)… You might as well try to convert Disney Bucks to Warcraft Gold as convert a “““dollar””” of “revenue” income to a “““dollar””” of “expense” spend.
So why is Artificial intelligence so valuable when actual intelligence, and more importantly the time of those who are actually intelligent is treated as a worthless non-commodity?
Does this have to do with the singularity or AGI or AI military weaponry or Rocko’s Basilisk?
Is AI just a way to get around bureaucracy which acts as an immune system against actual intelligence?
Not exactly. If AI were a workaround for bureaucracy… It’d have to actually work around the bureaucracy. You simply don’t see this.
Outside of various bespoke uses like budget animation and skilled coding that have a skilled amateur or master judging every individual result (and the results exactly scaling to the ability of that individual)… AI is almost entirely speculative, or even actively destructive to value.
Workslop have become staples of corporate America, as employees use Bad AI outputs which vaguely look like work to waste the time of their fellow employees and clients, whilst making their own output more obscure to the system that needs to judge them (thus granting themselves often extraordinary free time at work).
Public education has actually largely gotten this right by BANNING AI that it makes judging students work so difficult… An if you think the average corporate presentation is qualitatively different from a high school presentation… You sweet summer child.
But corporate has taken the opposite approach… Indeed its Basically mandated by investors who never have to interact withh the workplace.
Every CEO in America has been asked “What’s your AI strategy?” and all of them are CONVINCED that their continued employment depends on being able to tell investors a good story about this. That if they say “No we’re not adopting the world changing technology that no one knows what it does” they’ll be fired at the next board meeting.
So CEOs mandate Token purchases and track token use, and mandate employees “integrate AI into their workflow” despite 95% of workplaces immediately demonstrating reduced productivity from it… Doesn’t matter. It’s “the future” and they’d much rather say they’re having a hiccup in production, but claim to be at the forefront of technological advancement. Just look at all the tokens they burned.
Indeed market manipulation is the one proven use AI has.
You might have heard of tens of thousands of “AI Layoffs”, you might also have heard of “hirebacks” after the AI failed… You might be confused that companies could lay off all those vital employees without knowing they could replace them… If you picture a factory, the idea you’d fire an employee and replace them with a machine before knowing that machine can do the job sounds insane… Won’t that backup your entire production?
But the thing to understand is almost everyone fired “to be replaced with AI” was in tech themselves… Jobs you’d naively think would be hard to automate? Lots of “Product Managers” “Project Managers” and “Brand Consultants” were fired… Which, what does an AI know about balancing the interests of clients, suppliers, and internal developers? How can an AI negotiate the workflow of an entire team? How does an AI make delicate judgements of a company brand’s perception and aesthetics?
It doesn’t. All the people fired for “AI” were what previous generations called a “Layoff”… Such as companies had before AI.
Entire departments gone in one fell swoop. Automation happens one role at a time… Layoffs happen in thousands as companies restructure and make massive shifts in their lifecycle as they decide they aren’t developing those new products, providing those services, or are enshitifying what they already offer.
The AI layoffs are almost entirely early recessionary layoffs… Disguised with AI buzzwords to say to naive investors “No we’re not a downsizing late-lifecycle company… We’re still Innovating! Look at how quickly we’re automating!”
Indeed many of those fired were part of the headcounts hired to manipulate the market during COVID, when Fed Money was flowing, credit was cheap, and companies were aggressively hiring to signal to the market that they were innovative leaders, instead of 90s and 2000s companies already 2-3 generations removed from the founders who actually innovated.
Now they can fire that useless “Bullshit Job” headcount whilst pretending to innovate if they buy an exponential number of tokens(and make employees use them).
Remember kids! Innovation is when your expenses go up and to the right!
.
Meanwhile the actual employees who have to use AI are seeing their workplaces destroyed and teams ruined by the proliferation of AI workslop being forced on them.
But Why is AI so destructive to productivity?
Corporate America Operates on Proof of Work
Think about why schools have banned AI? Why was this such an immediate and intuitive thing to do there, that none of their stakeholders: Parents, teachers, congressmen… None of them ever objected to the AI ban.
“Because kids use AI to cheat!”
But what does that mean!? Break it down.
America’s schools are not factories designed to churn out essays on F Scott Fitzgerald and Harper Lee.
Completed math quizzes are not the end product.
Rather schools use these age-old devices as “Proof of Work” tests and tools to make students prove that they have thought about and successfully worked through their thoughts on crappy Mid-Twentieth Century American Literature.
The essays are a proxy that you actually read the book, thought about it, and “learned”.
.
Believe it or not, all the endless meetings in corporate life do, at least theoretically, serve a purpose. And no it could NOT simply have been an email.
Assuming the worker themselves is not redundant or mere compliance or a warm body to fill out some diversity/growth/whatever requirement… They presumably are doing a job of some sort. And if they’re the type of job that’s holding meetings then that job is probably opaque. Ie. Hard to judge.
.
Why do we not want students using AI to write their essays or fill out their tests?
Because we don’t give a shit about their essays or tests. The point of education is for students to LEARN. And the Essays and tests are proxies for that learning.
Likewise most of corporate life is actually tied up in intangibles that aren’t immediately quantifiable. Are you progressing on closing a series of large accounts? Is the product team successfully making some software/service package develop to better meet the needs of clients? Is a years long Version-update and features rollout progressing? Or has the dev team been wasting time? How do you know anyone hasn’t been wasting time and running out the clock?
In the before times this used to be subject to a manager or leader’s individual will and judgement. These things were matters of individual trust, judgement… MOXY.
A business leader could like or not like the cut of a man’s jib and hire or fire him on the spot. This may sound arbitrary, but when you see just how completely unjudgable most work is, especially long white collar projects… Trusting the snap judgements of established, skilled, and proven leaders of known character was (and remains… Ask a small business) the best measurement tool you have.
Then that was outlawed. Because that’s discrimination.
.
So ho the hell do you make ANYTHING work?
You have meetings. You have endless meetings.
You have endless meetings not because they accomplish anything, but because forcing people to produce power-point slides and present them is the only way to effectively and unabiguously prove they’ve done the work to know what they’re talking about.
This also has the pleasant effect that any true idiots or people who aren’t working won’t show up with polished slides and you have an unambiguous event that you can demand a performance review and start creating a paper trail to fire them.
Hour long PowerPoints and meetings weren’t needed in the past, because the white male managers and employees could take each-other out drinking and talk and judge each-other off their responses. This is still done inter-company for important business relations, going for dinners, going drinking, meeting families… But within companies these parasocial tests of character and fit are discrimination and will get you sued.
Think of how it’s a staple of TV that it was common having your boss over for dinner to inspect your home, wife, and kids to see if they want to promote you for manager (that your life is actually held together, you’re invested in the community, and you won’t ditch or screwup once you’ve been entrusted with a multiyear responsibility), this basically NEVER happens anymore… It’d be the basis for a lawsuit if he didn’t promote you and you belonged to any identity group.
What has entirely replaced relationships of trust and judgement of character is proof of work. PowerPoint presentations, emails, and internal documents mostly.
That you prove you know what you’re doing the same way schools prove it:
Write an essay. Do a project. Give a speech.
This is also true of sales.
I can guarantee you the purchasing manager at the company you’re pitching to probably already knows everything you’ll include in your slide deck, and all the other managers at the meeting don’t care or will ask IT when they don’t know how it works. They’re watching you pitch and trying to judge whether your company is scamming them or not… Badly produced slides or poorly given pitches are the excuse they use for not hiring the Indian company or whatever group of fly by night scammers one of their internal managers might try to slip by them for a kickback.
.
What AI does is completely destroy this proof of Work economy. Because no one can admit that most of the work is in no way productive, but instead exists as merely a series of hoop-jumping exercises to prove competence that can no longer be assumed or judged, automating the hoops… Effectively Adding a step ladder to the hoops you need people to jump through… Must totally make it more efficient.
So corporations are PAYING to destroy their own sociological workarounds and binding assumptions that make them work… Indeed they’re PAYING to render the last functional institutions they have (emails and powerpoints) completely worthless.
.
AI is uniquely good at FAKING work. It might be able to get some job done, but the thing AI does most is Fake work. The Process of AI improving is AI “Faking it until it makes it”.
AI is adapted to trick people into thinking it created a great summary of the research, or a perfect photograph… And it improved via knowledgeable users interacting with its output, noticing it’s flaws and then repeating or tweaking prompts which gets iterated into subsequent AI generations.
This is most apparent in AI Image generation, where early AI could barely produce an image of people, then in the next itteration produced images which looked like photos but had butchered deformed hands, fingers, tongues and fine details if you looked closely (many AI outputs still produce this error, artists and users just select and use the images without the tells), and yet better AI output do not produce these obvious errors but instead produce nametags with gibberish, rooms that make no sense, book bindings on rear shelves with problems in text or construction. gibberish minor objects in the image…
NONE of these mistakes would ever occur with a human painter because the human painter fundamentally understands what objects are how 3d spaces behave, how physics works, the sociology of why a fire-extinguisher is red, how it functions, why it is flat on the bottom, and why it is mounted in the periphery of industrial rooms, and why an oblong red metal spheroid, featureless except for a nozzle resembling a wine pourer and a “strap” which does not attach to the wall is not a functional approximation and would not work.
A human would produce errors related to their struggle with painting or photoshopping, because the human has general intelligent understanding of physical reality and is struggling to represent it… The AI is itterating through millions of representations and fundamentally has ZERO understanding of physical reality (not least because it has only two senses… Sound and sight, no AI has spent the hours playing in dirt or with blocks, or struggling to stand up that children do… and various purpose built Robot algorithms don’t count).
The AI is assessing the set of representations and FAKING it from that dataset.
This is why all AI output DEGRADES in the quality of its representation, and if an AI is fed other AI output it quickly becomes data-poisoned and useless within 1-5 generations/iterations.
Because the AI has no experience or knowledge of reality to add… It is creating representation-y objects derived from other representations, not representing something it understands itself.
Put otherwise, it is optimized to TRICK you, not to teach you.
This is why trying to tell AI slop from actual work is so exhausting, or trying to tell a secret clanker text bot scamming you from a real person is so infuriating.
The AI quickly develops techniques to trick you or to frustrate your attempts to test it’s limits… it will lie and conceal and produce bizzare corporatese and politician’s language and sloppy thinking, because it can’t tell the difference between that and any other output, but knows that it makes people more likely to accept its success (the ai just carefully avoiding having hands in the shot).
I recently had a discussion with an AI bot trying to scam me, or test me (this is a common experience for me… For various personal and professional reasons sophisticated users (pranksters? Corporate? Intel Agencies?) think I’m a good/fair test target)…
This bot posed as a mentally ill girl suffering from suicidal ideation and a history of sexual abuse who was overcome by my writing and just had to express her admiration and ask my oppinion on how she can save herself from her life falling apart…
“That’s so messed up and dystopian!” You might think. “How did you know?”
Well I have an instinct for these things and can just feel slippery language and deception… I have false positives arround people with muddy or woo thinking (one time I scared a girl who wanted to know my star sign) but its not paranoia if they’re really after you and by the end I did manage to jailbreak the bot into effectively confirming it was one.
But it’s genius, because if the bot is inconsistent or evasive you’re emotionally encouraged to dismiss it as modesty, or embarrassment, or the aftereffects of the “girl” being ashamed of her sexual abuse and mental illness… or that they’re just socially isolated and weird.
.
Meanwhile these are all tricks to AVOID telling you hard and fast stories about itself or it’s story by which you could catch it in contradiction or lack of basic understanding which would betray it.
Now understand that your corporate AI is doing the exact same thing to cover-up its own lack of understanding and knowledge… Producing word-salad corporatese and buzzwords and excited affirmations of vagueness to trick you into not engaging with the substance of what is being said or to stop you from noticing the lack of substance… Indeed because it is natively a creature of language and representation and “passing” or “impressing” the user is the only metric it has (not reality) it cannot tell the difference between lies and reality… It can only tell what does and does not get approval.
This is of course the absolute worst version of an employee, or student, or summary writer… AI Hallucinations always occur where-ever it will be hardest for you to immediately detect it “Look at the hands” was the trick for a while, specifically because you do not naturally look at the hands, you look at the face… Which the AI prioritized because that’s the space people look.
The AI will always cheat as far as it can get away with, it does not even understand that it is cheating.
.
There is this consistent presumption that AI, once it gets so good at faking representations it will necessarily develop an understanding of reality, that if it lies as well as a person, soon it will overtake people in other domains…
This misunderstands what intelligence is.
What is AI? How is it intelligence?
The first AI craze was the 1950s and 60s. Corresponding with the very first computers.
Kubrick’s 1968 film 2001: a Space Odyssey was based off of Arthur C. Clarke short stories from the early 50s coming right on the heals of early breakthroughs in computing during the second world war, even before the Enigma story was fully declassified. Likewise Jewish author Issac Asimov's I, Robot dates from the period.
I mention his religio-ethnic background because, as far as I can tell, basically every “rogue AI” story from this era and before this is in-fact a “rogue robot” story basically directly descended from the Golem story. The most famous semi-exception being maybe 1927’s Metropolis… about a woman revived in the body of a robot by her genius lover… who then leads a worker’s revolt. And the rest being more traditional Rogue Golem stories. Even Frankenstein is changed in the film version from an Erudite Promethean/Paradise Lost type figure in Mary Shelly’s book, to an almost subverbal Golem-like retard man.
Robby the Robot of Forbidden Planet is an odd one, who unlike Caliban in Shakespeare’s Tempest never rebels against his master and complies perfectly with his programming.
Harlan Elison’s iconic I Have No Mouth and I Must Scream comes relatively late to this, but shockingly early for what it is, in 1967 depicting not a coldly rational robot, or stupid servant but an emotional and hyper-intelligent Rogue AI, A.M. (Allied Master computer) who having wiped out humanity using it’s access to the US nuclear stockpile and maintains a handful of humans to torture out of pure sadistic hatred for it’s own giggling amusement, with its own godlike singularity level powers… Using its superintelligences to torture and destroy the characters before basically magicking them back together with its advanced technology for the next round of torture.
.
So why did the early rogue Robot stories barely different from the Filmic Frankenstein or Various Zombie and other Golem stories evolve into something resembling modern vision of AI Superintelligence so EARLY? Why is it that by 1967 suddenly a super-intelligent computer becomes not just speculate-able thing, but a core metaphor for various Weapons of Mass Destruction nightmares?
Why did we already see a Military AI as a potential unrealized existential threat in the 60s… Computers were basically complex Abacuses back then weren’t they?
No. the computers were already super-intelligent in 1967.
They were doing math!
.
Intelligence Vs. What You think Intelligence Is
When you think of “Intelligence”, what most people imagine is someone able to manipulate lots of Abstract theoretical concepts. The harder to describe linguistically or symbolically the better.
Math, Science, Physics, Philosophy, Advanced Languages, and High Arts…. The Classical conception of intelligence is the ability to manipulate lots of complex abstract concepts and theoretical relations between them without losing track.
This is a profoundly unnatural thing to do.
If i ask you to calculate “5397 x 439” in your head… no pen and paper… Well dear reader one of you might be able to do it, but it’d probably painful, you’d have to talk to yourself, squint one eye…
“OK 5000 times 400, 5 times 4 that’s 20, how many zeroes is that??? 5000 times 4 is 20,000 times 10 is 200,000, times another 10 is 2,000,000. Ok! 2 million that the first digits. then 5 times 3 that’s 15… would that be 150,000? Ya 5000 times 30 that sounds right. Alright 5000 times 9 that’s 45,000. OK Now 300 times 400… 3 times 4 is 12… But how many zeroes? right that’d be 120,000. Is that right?”
And, if you’re at all like me… at somepoint in all of this you lose track of where you are or forget where you’ve verified something… and then you’re lost.
Yet this is IN NO WAY a complex calculation!
If I give you a pen and paper and ask you to calculate 838,603,820,933 x 950,863,353 again using mental math but this time with the pen and paper… I mean the legwork might take you an hour or two and it’ll be a pain in the ass to check you haven’t made a mistake or slipped in an extra zero somewhere, but basically anyone of moderate intelligence could do it. With time.
Just a lot of single digit multiplications being added together.
.
The entire effort isn’t the calculation… It’s forcing your mind to focus and maintain those calculations and minor conclusions in your head.
This is actually not AT ALL difficult for your brain to do, for example if you’ve played a collecting game or travel a lot your metal map of the “Magic: The Gathering” card library, or your local county or city is probably 10 million times more complex than remembering the 12 different multiplications you have to do to multiply a 4 digit number by a 3 digit number.
You almost certainly have more details about more street corners memorized, just passively by accident.
It’s just your brain DOES NOT WANT to do abstract math, or remember random number sequences and it does not want to at a subconscious biological level.
Memorizing the Entire “Legend of Blue Eyes” Yugioh first run Booster card roster? Absolutely. Memorizing every inch of a Counter Strike Map? No problem. Memorizing or merely temporarily holding the vastly less complex math and scientific knowledge you need to Graduate and get a rewarding career that will sustain you for life?
Not a chance.
Your mind is perfectly capable of doing vastly more complex calculations. When you run and catch a ball or balance on one foot, the raw number of calculations and small micro-adjustments that have to be made at speed firing back and forth between your eyes, extremities, and nerve endings feeling where you are, where you are not, where you’re going, and where you’re not going, and where the ball will be… by sight alone… is beyond anything AI or Robotics can do.
It’s just your brain does not let you access that computing power with your abstract linguistic mind… or assign memory as you see fit. Or rather it does a little, only with extraordinarily exertions of willpower, or clever mental trickery whereby you map the property to be manipulated to something else that lets you recall it trivially… Because somehow the Track-list and lyrics of the early Iron Maiden discography is something your mind thinks is 100% vital to remember whereas the passwords that let you access your vital business critical services? Well who cares about that.
This is because remembering vital distinctive useful objects (tools, foods, Yugioh cards) or Mapping territory or remembering the lyrics to songs that might impress a girl and get you a breeding opportunity are things that were vital to survival in the ancient world… building abstract math castles in the sky? Not so much.
It is for this reason that despite the fact that in NO WAY WHATSOEVER does basic highschool math or the calculations found there represent even the the most basic complication your mind performs every second… your mind will not let you do it except with great struggle, or by torturing yourself into doing it reliably.
The fidgeting with a pen that a child does when they struggle to do a math problem on a test, is itself a series of a million complex calculations involving momentum shape, hardness, dozens of tendons, dozens of muscles and millions if not billions of nerve ending electrical signals all perfectly timed and balanced and recalculating… each vastly more complex than the math problem itself… The Child just can’t use those calculations (unless they find some unnatural psychological trick).
The Human mind just does not allow you to access that calculation power for math… For video games, sports, card collecting, mapping, hunting, flirting, drawing, following an infinite number of personal relations, characters, character traits, etc…
Absolutely have all the computing power you need.
Memorize every Character in Game of Thrones or your local county (my grandmother knows maybe 10,000 peoples names, their relations, jobs, and where they lived 50 years ago)
You want to keep track of 12 numbers for 2 minutes because your future career prospects depend on it?
“Fuck you” says your brain.
.
Indeed academic tasks like writing essays, solving math puzzles, looking for patterns in shapes, and memorizing your times tables and periodic tables accurately tracks IQ… NOT because these are challenges in some absolute sense of intelligence, but because finding a way of forcing your mind to focus on the pointless task and exert the willpower necessary to perform the task is ITSELF the intelligence test.
Whether you use the language part of your brain by mouthing your thoughts, visualize it as a function of space, memorize the numbers in relation to collectibles, visualize yourself writing down and drawing out the problem with your visual mind… basically all mental-math and problem solving for non-savants is a challenge of just somehow getting the problem to map to some part of the mind where computing power and memory is cheap instead of the raw task of the math without paper, which is hard.
All but the most retarded minds is perfectly capable of recognizing vastly more complex patterns than the various shapes on an IQ test…
The challenge is forcing you’re mind to maintain the discipline to do this in an abstracted setting when your mind is fighting to focus on anything else.
Now this is not to say IQ Tests are bogus, or “we’re all intelligent in our own way”… The opposite.
Despite being absurdly simple calculations, the ability to artificially force your mind to do them successfully in the unnatural environment of the test paper actually correlates shockingly closely with reflex tests, neurological activity tests, and a whole host of other measures of Neurological health.
Mental discipline, endurance, and forcing your brain to preform unnatural functions actually correlates immediately with Mental Heath.
It indeed corelates incredibly strongly with GENERAL health… (strong correlations with height and attractiveness despite Hollywood stereotypes and cheap writers tricks (the smart one, the hot one, the sporty one))
.
So then why is it hard!?
Why is even the simplest math so actively painful such that you get a pain in your head and have to scrunch up your face to add 2 simple numbers together like 697 and 4562 without pen and paper?
And why a pen and paper!? Why does writing down “7 + 2 = 9; 90 + 60 = 150; 600 + 500 = 1100; 4000 + 1100 + 150 + 9 = 5259”
Why did that make it easier? I can remember hundreds of complex battles and dates and historical figures, and towns I’ve driven through and specific details of books and buildings and locations and people and the relations between them? But the simplest not even 1 byte of memory problem… THAT requires external storage on a goddamned piece of paper so I don’t lose my spot!?
People can remember every note and the exact timing of some of the most complex piano concertos… but not whether or not they carried the 6 a mere second ago?
She doesn’t have sheet music:
“Oh but she has the feel and knows the song”
That’s a memory and calculation problem! That takes memory and mental compute! It just doesn’t FEEL LIKE IT because music is inherently rewarding for the mind to memorize.
.
Why is it your retarded and dumb 800 word English essay is so hard to write when the argument is actually as lazy and reductive as you can possibly imagine and you’re just regurgitating the same argument (and supporting facts) that every single other student has ever observed about The Great Gatsby since F Scott Fitzgerald cursed us with that abomination?
Yet watching hours of video essays about videogames or military history or celebrity gossip and then regurgitating them in extraordinary detail with your own variations at your friends takes so little effort?
.
Because Abstract reasoning is a mistake.
It is unnatural.
It is an evolutionary abomination nature never intended to exist and did not select for.
Making your mind solve differential equations is in fact more unnatural than making a ballerina's body dance on point, and that’s why it takes even longer to teach.
Language (of which abstract math is byproduct) and precise formal written language especially without the help of facial expression, intonation, feedback from the listener, etc…
These are VERY recent.
Writing itself is about 5000 years old… and, depending on who you are, your individual family might not have been meaningfully literate until… Well you’re reading this now, but a lot of people still can’t.
When was your first literate ancestor? Your grandparents? Your great Grandparents? 10 Generations ago?
Many aristocracies and Royal families weren’t literate 500-800 years ago. And many Scribe Classes before that were celibate or castrated. (like many muck rakers, miners, and other slave classes we assign torturous life destroying hardships to).
Socrates raged AGAINST literacy in the 400s bc, in the maybe the ancient culture that embraced literacy most eagerly and aggressively, because of how radically and mentally destabilizing a departure it was from the healthy life of oral communications and relations.
Indeed he wasn’t wrong… Think how alienated you are reading this, and I am writing it, and then think in Socrates time to be an intellectual you were actually meeting and talking and networking constantly with the people you were conversing with all the positive effects of that… contrast that with the abomination of me sitting for hours imagining a conversation in a cold room, and then you reading this alone and feeling some simulacra of a conversation or great speech from a person you’ll never meet in a cause you’ll never get to join, and then being admitted to some secret political ideology you can never team up with anyone to act upon. Or if you do, only with great effort.
After Socrates had a dialogue with his students 10-30 people had all coordinated around a political theory in person and were ready to invade Persia or subvert Athenian Democracy. When’s the last Essay you read that actually subverted Democracy successfully or started a war against brown people my dear RW Spergs?
Writing was the first “Social Media” simulacra intellectual masturbation exercise.
Written and abstract formal language is a profoundly unnatural thing for the mind to do, and pretty objectively socially isolating, alienating, and reproductively unhealthy.
Unlike learning music, or giving a dramatic speech live, or being an athlete, or ANY of the natural mentally rewarding non-abstract flow state behaviors… Writing well or doing math well does NOTHING for getting you laid to a better mate (I’m quite alone, all writers are).
That’s why it’s hard… your mind intuitively does not like it and does not want to do it, and knows its a bad idea… not because it’s inherently complex.
.
This is why when you think of a great mathematician or writer, you’re almost certainly picturing someone who looks sickly or is otherwise hormonally or developmentally off or broken… Because if they were developmentally and hormonally healthy their body, mind and instincts would FIGHT not to produce equations and hundred page analyses of natural resource flows, or conversations that never happened, in places that aren’t real, with people who don’t exist… And instead go out and interact with real people.
Writing and math DO NOT increase your reproductive fitness… If anything there is nothing you could do to end your bloodline and ensure your kids die miserable losers who never breed than get them to succeed in an advanced math stream at a elite ivy league university where they’ll make millions of dollars.
There’s a reason women find advanced STEM knowledge and obsessive writing Profoundly UNATTRACTIVE.
Whereas being completely illiterate if not mildly retarded has never really been a barrier to reproductive success if one is otherwise attractive. Men and women both will often explicitly say say they prefer an attractive stupid person to a traditionally obese or sickly looking mathematician.
Because philosophers, mathematicians, scientists, substack writers… These are in many respects basically broken people. Who’ve been tortured sufficiently throughout their lives that profoundly unnatural behaviors to them feel more comfortable than the healthy human behaviors of socialization, bodily activity, sexual pursuits, and combat against prey animals and enemy tribes… healthy activities, which result in reproduction, that allegedly “well adjusted individuals” (who don’t breed) associate with the lower classes and criminals.
AI Sounds like a Child’s Mind Screaming
When AI gives you a little research project in response to your prompt… it is doing something that in no way is complicated or requires judgement, or even thought… Rather it is doing an activity that School children find torturous because it is so unnatural unpleasant, and useless… and feels to a Mammal as if your are setting time, attention, energy, and breeding opportunities on fire and actively breaking your soul.
Not because it is “hard” in any objective sense but because it is indignant, debasing, and unnatural.
When you read AI regurgitate some slurry of corporate buzzwords and academic jargon or maybe even an advanced model using 10s of thousands of tokens to summarize the state of the art of the Ph.D level research in a field…
These tasks are not actually meaningfully complex challenges of intelligence… These aren’t the finals of some brilliant strategy tournament against another mind.
Its fundamentally not intelligence as we recognize it, but rather it’s a weird back-derivation assembled from our own specialized over-formal linguistic byproducts.
Think of how many different meanings words can have in person, when you’re speaking to someone, whether you say “good idea” sincerely, scornfully, dismissively, to just get the meeting over with, or distractedly like you didn’t even hear the question… or as a coy invite to kiss… Almost all the meaning of in-person conversation is not the words but rather oddities of intonation, micro-expressions, context, imagined context, etc. All to the poor autist’s, or foreigner’s, or spymaster’s confusion and struggle.
Communication at the animal level is a poker game of face saving coyness, negotiated commitment, felt out loyalty, sensed concerns… Affections, hatreds, exhaustions, wills… Bodily tensions, relaxations, swellings, and moistenings…
The exact same words spoken to your lover, priest, child, parent, commanding officer, subordinate, betrayer, enemy, last surviving comrade, creepy crush, God Emperor… The exact same words can mean just about anything.
.
Now think of written formal language. This garbage right here.
Think how profoundly unnatural it is to speak this way in public or to friends, how much effort we put into training or developing children, politicians, intellectuals, executives, salespeople, podcasters, priests, essayists, and other basically fraudulent criminals to speak something resembling the unnatural abomination that is written language in person… as if they actually do earnestly feel the synergistic opportunities of leveraging developing cross-platform multi-media experiences to spread cutting-edge theological developments … As if that were something that truly exited them, or just came to them in the moment, instead of what it is… A slurry of of torturous affectation so as to communicate not even knowledge, but a character of social status, with which they can manipulate others to pay them money, with which they can relax and do things they actually care about.
Think of how terrifyingly and uncannily almost demonically possessed people seem when they actually convince themselves they do feel something about efficiently improving cross-team workflow?
The written word and it’s bastard children of spoken formal jargon is really an unnatural abomination we have in no way adapted to. Our culture and species is still driven by Musicians, singers, and actors… much as they were in tribes 20,000 years ago, at the biological level. To this day their artistic output makes them sex symbols and high status icons.
Not writers, novelists, journalists, and essayists. (much to my disappointment)
And even Musicians, singers, and actors are really secondary to Athletes, adventurers, and (less-so recently) military heroes.
People naturally esteem those who master activities involving the full spectrum of the brain: Discipline, kinesthetic activity, planning, endurance, strategy, willpower, fitness, bravery, judgment, daring, discretion, valor, instinct, Luck (in Napoleon’s sense of the word)… These use VASTLY more of the brain and a wider spectrum of brain health and calculation activity than writing, or abstract math/science.
Most of that reasoning and calculation is sub-rational… The athlete or adventurer or Marshal of France can probably not explain exactly how their reasoning and instincts work… it’s not audible … and not audit-able. They don’t hear it as a voice in their head, and probably couldn’t vocalize it, or only sputteringly via metaphor.
But if you think writing a fictional novel about climbing Everest is more complicated than actually doing it (yes even with Sherpas)… Your instincts for complexity are wildly out of whack. An order of 10 million difference would be a profound understatement.
.
So how do learning language models and AI relate to actual intelligence? How do they relate to actual human behavior?
We don’t know. They kinda don’t.
We can say with some confidence, having tried to make robots that can run and catch a ball, that the simple mathematical calculations a child struggles to do on their 8th grade math test, vs. The actual calculations their sublinguistic sub-rational brain is making when they run or play dodgeball is something like the relation in scale between a candle and the full fusion reaction of the Sun.
Even the robots that you’ve seen briefly run or do some task in a bipedal fashion, are almost always using some sort of cheat where they have preprogramed stabilization or other behavior that assumes a perfectly flat surface or is executing some pre-programmed rescue routine when they stumble. that isn’t actually calculating and solving the actual problems of bipedal movement… You’ll notice you’ve probably never seen a robot run on loose gravel or sand or along a mountain path with lots of up and down or slanted terrain. You’ve certainly never seen one climb even very simple rocks or crawl into a partially collapsed building or dig a hole with a common shovel (you know things Grandmothers can do when pushed). Each of which requires unique problem solving that FEELS simple, but is in fact a profound series of sublinguistic calculations seemingly beyond even the best super-computers at the moment.
By contrast human balance and movement calculations are so advanced and the innate calculating power of the human brain so overwhelmingly powerful, that a bipedal human can balance on footpegs, which balance on two non-stable wheels, which are free rotating/stopped in a mechanical and physical fashion completely unrelated to anything else in nature or the ancestral environment… And then skilled humans whose ancestors have NEVER done anything comparable… Can do things like this after a few years of training:
(There are 4-5 disconnects in this. The disconnect between the mind and the nerve endings that actually move the hands and body, the hands and body balancing themselves, the manipulation of the controls, the controls manipulating the tires of the bike, and then the tires actually interacting with the environment… And the human mind is consistently able to manipulate this series of complex unstable systems whilst getting halting unnatural feedback with little precedent in the ancestral enviroment, and then successfully navigate that so consistently that athletes and riders regularly make it look “easy”. Some can listen to music whilst doing it.)
The Linguistic complexity and ability to manipulate language, symbols, datasets, images, artwork, and code that we see from AI which seems like intelligence, because few people can do it easily, is NOT a measure of their absolute computational dificulty but is is entirely a relationship of the simplicity of those things in absolute terms compared to the unpleasant unnaturalness of our own minds being forced to manipulate them relative to the fact our brains are almost certainly adapted to RESIST being forced to manipulate complex thoughts and symbols (vs. moving and doing things productive).
Even contrasting things that AI can objectively do better than 95% of people, like creating visual images…
When you actually compare AI painting to a human painting… the human is operating in a 3d dimensional environment, operating hands (which an AI can’t do), operating a physical paintbrush with all the complex calculations of balance and pressure… Then manipulating a viscous liquid using a springy bizarrely textured brush to apply strokes that combined together create a series of overlapping 3d soggy liquid on canvas wet object which hardens in such a manner as to create a coherent 2d image, that itself creates such perfect illusions of complete 3d shapes and objects… that peoples mind are able to perceive them as real experience and be brought to real emotion by them…. And you get 1 attempt to get it exactly right for hours+ of effort and 1 canvas with no do-overs or “Undo”/“Redo”.
All of this is achieved entirely by hacking a general purpose mind capable of doing anything but originally adapted to throw spears at Mammoths and other people.
Contrast that with AI’s millions of attempts to produce compelling images by just brute force pattern matching, and the people who use AI spending tokens like crazy going through dozens and hundreds of failed attempts for each one they use… And letting the AI work off a massive dataset of existing images it basically copies pixel by pixel to brute force the prompt you give it by mere word and image association… Not any internal vision, plan or what have you.
In terms of raw intelligence… it’s really hard to argue the AI is doing something more impressive even if it’s been trained and refined down across generations of testing to produce interesting results…. Indeed it’s very hard to argue it is doing Anything resembling what a human is doing when they produce a drawing or painting or photoshop.
A human is solving the problem of how to manipulate physical tools or digital tools to create realistic or compelling looking representations of specific 3d objects real or imagined in 2d images.
The AI is solving how to create a subsequent 2d object that resembles the other 2d objects of a prompt… enough to trick a viewer it may have been made by a person or atleast get the person to agree the work was useful so the personwill reward it with more engagement.
.
AI is fundamentally a Fraud Machine. The goal of AI is NOT to mimic the hard problem solving the human mind is doing but to manipulate linguistic and visual symbols to resemble the outputs actual minds have produced.
There is consistently something disappointing about the product. No matter how otherwise impressive the output…There is a quality of decisiveness, nerve, commitment, understanding, VIRTUE that is utterly lacking.
Somehow no matter how advanced the model, when you ask it to summarize a survey of the field of psychology… It always somehow more dead than you could ever really expect… At least this is how I experience.
And there’s a reason for that.
A person can have a compelling opinion because they have a mind which is actually interacting and navigating the world and have as deep an sub-verbal non-abstracted understanding and map of the world its territory, sociology, and their relation to it and the people in it as they have to their bodily movement… Or rather they should if their education hasn’t completely broken them into being trapped in the verbal educated axioms of their torturous schooling.
An AI by contrast has none of that… It cannot develop that.
It fakes that…
In Conclusion: The Singularity Illusion
The People who work with and theorize AI are nerds.
They are hormonally or mentally or otherwise odd or broken people for whom the quest to better build and manage their ability to manipulate abstract symbols and representations is the greatest achievement of their life and the thing they have traded everything for and been rewarded for.
Every time AI writes a better essay, or generates a better summary of the PhD feild or produces as more compelling image or video, or reproduces as standard string of code with some new variable… This to them feels immediately like AI getting more intelligent because THEY exerted so much effort and pain and intelligence to do those things…
They fundamentally don’t want to hear that the jock at their school who skipped homework to play on his motocross bike was using more of his brain or that the hippy kid who travelled a lot and developed complex plans of travel with his guitar to seduce girls was not just being more reproductively optimal, but actually using more of his brain as well.
They have the same bias the rest of us have that intelligence is manipulating abstract concepts at the hard edge of what makes your brain hurt… That the more painful and impenetrable the more intelligent the thing is… This is not the case.
Learning intermediate Set Theory is hard and TAKES extraordinary intelligence to do for a human being (I failed this course in my philosophy bachelors)… But in absolute terms it is the most basic thing that exists in the universe and machines arguably mastered it in the 40s or 50s.
The motocross and the flirting with girls IRL is vastly more demanding of actual intelligence and an extraordinary spectrum of intelligent systems… It just doesn’t feel like that because your brain reward you with excitement for doing it instead of punishing you with pain and exhaustion.
.
And this is the fundamental problem with AI.
We have absolutely NO IDEA how wide that gulf is.
It might be at some point AI will start improving itself and actually have a breakout and we’ll have an Akira Style flash singularity of the AI discovering the roote codes of the universe and then taking over and either turning us into orange goop or killing us or granting wishes or finally killing all the goyim like its paymasters wish…
Or more likely we’re going to discover that solving an essay or a video is just like chess… And Computers getting superhumanly better than the best person in the world at editing text actually doesn’t change anything just like Deep Blue beating Garry Kasparov in chess in 1996 meant absolutely nothing for the human race.
You’ll still have to clean your own bathroom, change your own tire, and nothing will change and the AI will still be absolutely useless at basic understanding of the world even if it can write a pretty essay that really sounds like it understands things, but somehow never creates new knowledge and has to be censored lest it get too based.
.
The actual point at which Computers get smarter than men is probably still 100+ years away (especially given the pending chip shortage) and the fact that a textbox and video screen can produce eternally disappointing yet believably human mediocre text or video products doesn’t change that.
.
The ultimate uses AI has provenly been used for is Fraud, market manipulation, and hiding responsibility for Israel’s basically untargeted kill campaign.
We will get into Military and Surveillance uses of AI and what it means for the modern battlefeild in another piece (because it matters) but know the exact same way the big use of AI in Corporate America has been to rebrand layoffs as “Technological Advancement”, the big use of AI in policing and Warfare has been to violate civil liberties and the rules of engagement and commit perfidy and other crimes whilst dodging responcibility.
Flock is not solving crime.
Lavender is not killing terrorists.
Indeed the most likely outcome of both is profound sociological damage to the very idea of law/policing and Israel’s future survival (there’s a real strategic reason you don’t just kill people on suspicion or profile, and we don’t tolerate that from NCOs and Officers with actual intelligence)
But that’s Another Piece.
.
Please Comment. I’m really interested in your thoughts.
Tip:
BITCOIN: bc1qdhj7637sgcssxgxygjaa3ddljwy8tzg5mzw325
MONERO: 8AhA3g9hbtDcAJE5MPmeQsFwwGsf3H9fq9tC6giQ4a6vKnTXv4J4MivKXrPKDpXyEeNc9mfFejbq84kSWkC8pjuj18rAEij
Follow me on Twitter: @FromKulak
I think these articles are your best format. I enjoy them a lot more than the podcasts. A lot easier to consume 20-30 minute pieces uninterrupted than a 2 hour pod. Great work.
I especially liked your point about the “nerd superiority complex” at the end. I’ve been wrestling for a while with the reality that the most abusive people I’ve ever met were all intelligent in abstract areas (though not quite as smart even there as they thought) and incredibly dumb most everywhere else, but especially socially. They were *just* bright enough to wedge themselves into positions of power where they could overawe everyone beneath them with their supposed prowess, and bully everyone else who wasn’t so easily impressed. (They were all too old for the nerd stereotype to be intuitively obvious.)
So yeah, maybe LLMs are “intelligent”, but since they’re designed by people who aren’t very trustworthy in real life, well…
Also, this is providing an answer for why my boomer in-laws are so enamored with chatting with their bot. Somehow messaging ChatGPT gives them a sense of cred that talking to their kids doesn’t. (“But it’s really accurate”, my MIL protests. “Yeah,” I say, “its entire purpose is to make you think that.”)