The Cosmos Institute recently published a pretty successful essay titled You Can’t Speedrun Yourself.
The author Alex Chalmers argues that the difficult, tedious work of learning something is often inseparable from the learning itself. AI might give us the output of intelligence, but it cannot necessarily give us the judgment, taste, and understanding that comes from years of disciplined practice.
If you automate the drudgery, you might accidentally automate the very thing that makes you good at what you do.
With all due respect, I think Alex’s argument is wildly incorrect.
Progress is the point
First, we must establish some basic principles. Assuming the goal is human progress, we ought to pursue the fastest path to it. This seems uncontroversial to me.
Every generation inherits a world that was painstakingly constructed by the generations before it. We no longer have to discover fire, invent calculus, or figure out how electricity works. These things are handed to us as the baseline conditions of our existence. And what an extraordinary gift that is!
A reasonably curious teenager today has access to more knowledge than the greatest intellectuals of antiquity could have dreamed of. We are truly at the beginning of infinity. Knowledge creates the conditions for more knowledge, which creates the conditions for still more knowledge.
Chalmers cites the philosopher Albert Borgmann, who laments that central heating replaced the family stove, thereby relieving us of the rituals of chopping wood, tending the fire, and paying attention to the seasons. Something meaningful was lost in this transition, the argument goes. We got warmth, but lost our relationship with the process of producing it.
I imagine Alex wrote his essay in the comfort of a centrally heated home in the UK. Instead of spending an afternoon chopping wood and tending a stove, he was free to write an essay. His own argument is, in a very literal sense, made possible by the thing Borgmann lamented.
Chopping wood can be pleasurable, meditative, physically rewarding, etc. So can knitting a sweater, tending a garden, or making pasta by hand. But I would much rather live in a world where these are things we choose to do recreationally, not things we must do.
The trouble with the monk analogy
The essay opens with a fourth-century Greek ascetic, Evagrius Ponticus, who described the so-called noonday demon. The demon tempted monks to abandon the difficult path for something easier.
Alex uses this as an analogy for the age of AI: Just as the monks had to resist the temptation to abandon their duties, we ought to resist the temptation to outsource the difficult work of intellectual formation.
I see why this framing would do well on Substack, but again, wildly incorrect.
For one, the monks were operating within a fundamentally different conception of human flourishing. Their goal was not to push the frontier. It was spiritual salvation, pursued through obedience, renunciation, and dogmatism. Curiosity itself was regarded as a distraction from higher spiritual duties.
Side note: Contemplative traditions are useful in that they have sophisticated techniques to understand and silence the mind. Some of it might turn out to be extraordinarily valuable to modern neuroscience. But I do not think a monk trying to resist the temptation to be curious offers a good blueprint for human progress.
The essay concludes with an anecdote about Abbot Paul (gosh what is with these anecdotes!) who spent his days weaving baskets from palm leaves, only to burn them at the end of the year. His labor was valuable precisely because it produced nothing. The purpose was the discipline of doing it.
Well, godspeed Abbot Paul. But imagine extending this philosophy to the whole of civilization. Imagine insisting that we continue performing obsolete intellectual tasks because the performance itself builds character. We would have a tremendous number of very disciplined people, but I’m not sure we would have very much progress.
The absurdity of abundance
For most of human history, intelligence has been scarce, expensive, and unevenly distributed. We’re now making significant portions of that intelligence available on demand, at a rapidly declining cost.
I think we’re still too preoccupied with the immediate implications to appreciate the strangeness of what is to come.
We’ve seen this before with calories.
Getting enough calories was the central problem of human existence. Enormous amounts of human labor were organized around producing, preserving, and distributing food. Then, through agricultural and technological progress, calories became dramatically cheaper.
This was an enormous achievement, but it also produced some rather strange outcomes.
We developed new problems associated with overconsumption. We invented diets to restrict the very thing our ancestors struggled to obtain. We built gyms full of machines that allow us to burn calories without accomplishing any useful physical work.
This is the sort of absurdity that abundance creates. Imagine explaining a treadmill to a medieval peasant.
Cheap calories changed our relationship with physical effort. I suspect something similar will happen with intelligence.
Evaluating new technology based on old constraints is therefore incorrect. Our institutions are full of practices that exist because of these old constraints. We confuse these constraints with virtues because they’ve been around for so long.
For instance, Chalmers worries that automating the work of apprenticeship might deprive us of the judgment that apprenticeship once produced.
But why should we assume that the historical method of acquiring judgment is the optimal one? The fact that intellectual accomplishment once required a particular kind of suffering does not make that suffering sacred.
Maybe AI will allow us to develop judgment faster, by exposing us to more examples, more counterarguments, more simulations, and more feedback than any traditional apprenticeship could provide. Maybe it won’t. But surely this is an empirical question, not something we can settle by consulting a fourth-century monk.
Perhaps philosophy needs a software update
This is a WIP thesis, so bear with me here. But I can’t help but wonder if philosophy is a bit low-tech for the current age.
Neuroscience, biology, math, CS, etc are all giving us new tools to understand the mind. We can study neural activity, model learning processes, investigate the biological basis of cognition, and build artificial systems that exhibit increasingly sophisticated forms of reasoning.
We need net new ideas for the future, not merely more elaborate ways of explaining how old ideas fit inside this new world.
Substack x Cottagecore - a strangely successful marriage
There is a specific formula of Substack essay that I find super difficult to tolerate. It goes something like this:
Open with an obscure philosopher from antiquity lamenting the human condition.
Explain the etymology of a Greek word nobody has used in 2,000 years.
Argue that modern technology is robbing us of some essential human quality.
Insert crackling hearth, medieval monastery, or craftsman who spent forty years making chairs.
Return to the philosopher. Imagine how horrified he would be by the invention of ChatGPT!
Add cover image of a classical artwork. (guilty!)
Publish to thunderous applause from people reading on their iPhones.
Anyone who has spent enough time on Substack can see why this works. It lends the argument a certain gravity, as though the mere act of quoting someone who lived before the invention of indoor plumbing confers intellectual seriousness.
Obviously there is a huge risk in this style of thinking. The past becomes a vast repository of anecdotes that can be arranged to support almost any conclusion. One can always find an ancient philosopher to explain why some modern development threatens the human spirit.
Personally, I’m more interested in the philosophers who haven’t been born yet. The people who will grow up with intelligence as an abundant resource, who will develop ways of thinking we can’t presently imagine, and who will look back at our contemporary anxieties with a mixture of curiosity and amusement.
I hope they do.