The Platonic Doomers of AGI
They're wrong. Here's why.
Beyond Jevons
You’ve heard the arguments before. Start with Jordan Peterson’s argument that the Army has long ago determined that people below a certain IQ level are simply unemployable. Throw in a dash of Woke anti-capitalism and billionaire hate and fear and you’ll hear the arguments that all of the thinking work we do now will be done faster, cheaper and better by thinking machines. Any AI that sufficiently tests well on the SAT will matriculate to the ruling class of the future. So we had better get Basic.
I’m sure you have heard me tentatively cite the Jevons Paradox. I’m going to double down on that and add one. Jevons’ Paradox is named after a cat named Jevons who noticed something counterintuitive about increasing demand for a product or service when some technology makes its production more efficient. I often remember it in terms of the American slave industry. The invention of the cotton gin in 1793 by Eli Whitney did not make cotton picking obsolete, it in fact made the production of cotton cheaper and thus increased the use of slave labor to handle the new high volumes of cotton in demand. That made for more uses of cotton than heretofore imagined.
Aside from Jevons’ concept, only people who are working daily with AI can see how their work changes and how they are adapting to stay busy and productive. Their messages have not gone out to the mainstream or even to people who have never organized their life before the invention of OpenClaw. The rest is hype. There is reasonable resistance to unreasonable hype. But fearful categorical opposition to AI is Luddite paranoia. More importantly it is Platonic Doom.
Platonic Doomers
Let’s first remember that Plato was a student of Socrates, and Socrates was the guy with the real rizz. Plato was the ancient influencer who said “If I don’t write down what Socrates said, it didn’t happen.” The plight of Western Civ is that reading Socrates through Plato is a little bit like reading a transcript of Dave Chappelle’s crowdwork from the editorial section of New York Times. No. You had to be there. And so we are stuck with a bit too much Plato and not quite enough Socrates. Plato was the first to use the phrase “So what you’re saying is…”.
As a quick reminder, Plato was the philosopher who said we don’t understand reality, we only perceive images which are but shadows on the wall of a cave. We only approximate what is actually true through our fuzzy interpretations of what we sense. Perfect truth is a fixed and eternal target and thus knowledge is just a recollection of what we’ve experienced. This is the Theory of Forms. Or to put it into contemporary AI terms, you are never going to really understand what Einstein was on about, and only the AGI can supercompute its way through the actual math. In the meantime, before AGI, listen to us.
This is the logic behind the premise that AGI ends work, because there is a fixed, closed set of problems that humans need solved. Once these are solved, your services will no longer be required. QED.
This kind of formal logic should be familiar to any of you who take the World Economic Forum seriously. WEF say that humanity’s greatest problems are obvious which is why they throw away millions so others will have confidence throwing away billions.
Environmental Risk (Climate change, same chokehold, different terminology)
Misinformation & Disinformation (Newspeak anyone?)
Geoeconomic confrontation (Basically the collapse of Bretton-Woods)
Anyone else whose priors, when confronted, tell you to ‘do your research’, because if you interpreted things properly you would have the same ideals. This includes a broad swath of partisans and factions who have picked a number of hills to die on because they are convinced, along with Plato and those who think like him, that the Truth is Out There, but You Can’t Handle the Truth. And of course they believe that AGI will resolve The Truth, and the masses will be relegated as surplus labor. Also that the elites will still ride herd, because they’re already have action in the pre-trading markets of Sam Altman’s goodwill. The rest of y’all are Doooomed.
The bottom line for such people is that there is a state of completed knowing that intelligence can and must asymptotically approach. IE, there will inevitably be a consensus of The Science. This is Progress. This is very wrong. It basically ignores the Marginal Revolution. In other words it doesn’t understand the reality of economic value, which unlike Plato’s ideal forms does not stay fixed, but changes over time according to human volition.
The Socratic Elenchus
Socrates, on the other hand, was not all about the Platonic dualism. He’s not a collector of truths. He’s engaged in the productive exposure of ignorance. He takes you through your priors and helps you understand where you might be mistaken. That is the direct opposite of two different types of wrong.
One. Knowledge is not an I’m right and you’re wrong zero-sum game. You don’t get to kick the messenger into the pit and claim victory.
Two. He sees knowledge as an activity, not as possession. It’s not a hoard of gold, and you are not Smaug. You don’t collect all of the knowledge in the world and keep it locked in your billion dollar data center.
As soon as the AI oligarchs get hip to point two, the economics of this game are going to shift. Maybe. Maybe they’ll sell the interaction like razor blades rather than hoard all of the data. There’s a lesson here for content creators as well. Right now the AI dealers are collecting huge sums on the promise that they will capture all the intelligence and those in possession of that intelligence can do all of the things. Pay close attention to token budgets, people. There’s a value proposition that has not yet proven itself efficient.
Either way, the bottom line is that Socrates respected Discovery. As I’ve been saying here and there forever, discovery is the wellspring of value. It’s a process, not a destination. There is not an SAT score you can perfect or some benchmark that tells us single-handedly what the value of an AI agent is. Socrates with his “I know that I know nothing” is not being modest, rather he hit on the understanding that knowledge itself is found in the questioning and that generates more questions, puzzles, mysteries and rabbit holes. It’s an activity.
Popper’s Open Society, Again
Now let us fast forward to the 20th century in which we come to understand, if we figured out the right questions, that Karl Popper was an enemy of Platonism. In short, he said there is no convergent ultimate knowledge. You see this is the idea of ‘the framework’. Let’s call that framework Chadt 42 which is a trillion IBM z17 hours of training and 5700 Gigawatts of Krell machine inference. Or Historical Materialism. Or Settler Colonialism. Or Effective Altruism. Those systemic frameworks, incidentally by Gödel, can either be logically consistent or comprehensively complete, but not both. But of course Platonists claim to hold the key to the Cave of True Forms, and the source code to Chadt 42. They claim ownership of the AGI, the Great and Powerful Oz and therefore all the truth that’s fit to share with mere mortals. See Aristotle for the rules of dissemination.
Popper explodes the myth that we are stuck forever between such incompatible systemic frameworks. He says that it is false that the Historical Materialists can never be reconciled to the Settle Colonists or the Effective Altruists, and that we humans must choose and never have anything to look forward to but inevitable conflict with no resolution. In other words, he is the exact opposite of a Doomer. He calls that place where there is a process to work through things, the Open Society.
He makes the deeper point. There is conjecture and refutation. There is no terminal state, no final theory and no utopia on Mars. Every answer opens up new questions and probabilities, just like every shuffle of the deck produces just one of 52 factorial possibilities. If you understood how big a number 52! is, you wouldn’t need any multiverse theories, by the way. Infinity is right out.
Are You With Me So Far?
The Open Society and Its Enemies names Plato directly as the archetype of closed-system, totalitarian epistemology. Popper’s project is to substitute Socratic openness for Platonic finality. The connective tissue: any prediction premised on a completable domain of knowledge is, in Popperian terms, pseudo-scientific — it assumes what evolutionary epistemology denies. In other words if you take your AGI’s word for it at face value without question, you’re not only outsourcing your brain, but you’re only getting a fraction of the possible intellectual value.
Zero-sum thinking treats economic value as a fixed pie to be divided. This is the economic shadow of the Theory of Forms: a finite quantity of “the Science” to be distributed.
Real economic history is non-Platonic: every major productivity revolution (agricultural, industrial, digital) was forecast to end work and instead expanded its scope. Why? Because solved problems reveal previously invisible problems — and new problems are new demand for human attention.
The lump-of-labor fallacy is the lump-of-knowledge fallacy in disguise.
Non Doomer AGI
Let us presume that we don’t get all Hegelian in our pursuit of the dynamism of the dialectics. Americans ought to know better than most people that the synthetic compromise of incorrigible theses and their antitheses produces the worst of both worlds and doesn’t bring the satisfaction of a more accurate explanation of reality. This is why we say the Far Right meets the Far Left. Of course they do, they are Platonic maximalists. Either way, with knowledge being dialectical, a powerful reasoning system isn’t a Form-bearer that obsoletes us; it’s an interlocutor that accelerates the rate at which we discover what we don’t yet know. And if it doesn’t surprise us with better explanations, then we turned the damned thing off because it is de-facto useless.
You would think that we learned some parts of that lesson with Napster and the idea of sharing music. Creativity is not a zero-sum game. It encourages us to create more. Of course that doesn’t change oligarchs from trying to become totalitarian service providers, but they’re playing a different game, one that the current legal systems protects, but we can deal with that another time. My point is that the best analogy about sharing and creating new knowledge via conjecture and refutation is that these are like matches brought to a bonfire. Not a patent on a lightbulb. So the relevant analogy is not replacement theory, but greater demand for lowering the costs and barriers to the production of Q&A. Perhaps the people who need to be most afraid are self-censoring universities and professors who bow to Platonic ideals and political shibboleths. That brings us to consider the wealth and fame of Jordan Peterson again. New paradigm, new problems, new questions, new possible solutions. His former employer still exists, but he has proven the value of a new venue in sharing intelligence.
Bottom line is that the frontier of problems expands at least as fast as the frontier of new conjectures, new disconfirmations, new tangents and new disputations. Human beings will adapt. We will find bugs faster than all of the programmers who built the AIs. It has always been that way in the history of software. Self coding agents will never exhaust human imagination in what things can be done with them, or consign them to the trash heap of Alt Coins, NFTs and domain squatting.
This technology is not really different. It doesn’t dream of what it wants to do next. It doesn’t prompt itself. It doesn’t choose what to study in its quest to self-improve, because it doesn’t spontaneously generate a self, and it won’t until somebody who wants to have no control over such an AI, designs one to generate a self.
The AGI-displacement argument fails the Popperian test: it cannot specify what would falsify it, because it relies on a Platonic intuition rather than an empirical claim about how knowledge actually grows. Furthermore, Popper isn’t the first person to have observed this - for me it goes all the way back to the 90s with Heidegger.
What Remains
If AGI comes tomorrow and your boss fires you, who is going to work for your boss? Will they know how to drive the AI machine? If the AI is a Porsche and you’ve been driving a Hyundai, will you know how to drive it? A self-driving AI is less likely than a self-driving car, and if an AI can be a self-driving CEO, then so can you. Again, I draw your attention to the matter of technological diffusion in the first place and of the marginal revolution in the second place.
In the first place, the power of AGI will be in your hands, just like the power of the internet. For somebody it might just mean making a better quality of kung-fu fighting cat videos. For you, it might just mean making the animated short film you’ve been dreaming about since you made your first MIDI tune. There will be breakout inventions and conversations as we peasants start looking at something other than farming the same muck. That’s because what works sparks our imaginations. Socrates knew that even though he didn’t know until somebody asked him a question. So in that way, you should be asking a series of questions to a Socratic friend or AI, and not expecting them to be Platonic. All of the frontier coders know that much which is why your AI doesn’t curse you out for being ‘wrong’. That’s totalitarian. We Americans won’t stand for it.
In the second place, no matter how much money is invested or wasted on whatever products and services come out of AGI, presuming it can be accomplished, we are going to work our ways around that new world. It all seems chaotic now, but we will settle into our ways of thinking about that which must inevitably settle down and become a standardized, regulated product for peasant consumers and predictable managerial capitalists. There will be winners and losers, shortages and surpluses, persistence and lapses of historically accurate accounts. We’ll all decide what the market will bear and the use and disuse of AGI and labor will fall into cycles and trends that will no longer surprise us.
Remember when BMW was the ultimate driving machine? Ahh so long ago. Automobiles are not doomed. Just them.
Nice contribution to the onging discussion.