There’s a scene early in John Kenneth Galbraith’s The Affluent Society where he describes Americans of an earlier era regarding industrial output with something close to reverence — the sheer productive capacity of the nation seemed almost miraculous, a force that could reshape civilization. Within a generation, of course, that same output had become background noise. Factories hummed, goods appeared, and nobody paused to marvel.
The miraculous had become mundane, and the mundane had become infrastructure.
I found myself thinking about that arc recently while listening to Sam Lessin on the More or Less podcast.
Lessin made an observation that I haven’t been able to shake: we probably aren’t heading toward a single, triumphant AGI monopoly — some god-machine that one fortunate company builds first and then rents to the rest of us in perpetuity.
Instead, Lessin suggested, we are barreling toward something far more ordinary, and in its ordinariness, far more interesting.
“There will be lots of ‘dealers of intelligence’. No one company will corner the market, no one big winner of AGI.”
Dealers of intelligence. I keep turning that phrase over. Where do we end up? No rapture, no singularity, no chosen company ascending to the throne of cognition. Just suppliers, distribution channels, price competition — the unglamorous mechanics of any maturing market.
And historically, that’s exactly how this tends to go.
Salt was once precious enough to pay soldiers with. Spices rewrote the map of the world. Steel, oil, and computing power each arrived wrapped in mystique and guarded behind scarcity before the inevitable happened: extraction improved, distribution scaled, and the miracle became a utility. Nobody thinks about the engineering marvel of the electrical grid when they flip a light switch. They just expect the light to come on.
If Lessin is right — and the competitive landscape of the last two years does little to argue against him — intelligence will follow the same curve. Not a single oracle, but a market. Cognitive utilities. Price-per-token negotiations. The same forces that commoditized bandwidth will commoditize reasoning, and we’ll argue about our AI subscription tiers the way we currently argue about our data plans.
Which forces the interesting question: when genius is cheap, what exactly becomes valuable?
The professional moats of the last century were largely built on the ability to process specialized information and output reliable answers.
The doctor, the lawyer, the financial analyst, the programmer — each occupied a protected position because access to their domain of reasoning was genuinely scarce.
If I can buy a substantial fraction of that reasoning from a commodity supplier for fractions of a cent, the premium on raw cognitive horsepower doesn’t just shrink. It collapses.
What’s left, I think, is the un-commoditizable. Empathy. Physical presence. Judgment under conditions of genuine uncertainty and consequence. And above all — taste.
Taste is the thing that has always resisted systematization, because taste isn’t rational in any clean sense. It’s the residue of lived experience, of specific childhoods and particular failures and the accumulated weight of caring about things over time.
An algorithm can produce a structurally flawless piece of music; it takes a human to decide whether it matters, and why, and to whom.
That act of curation — of choosing what deserves to exist and what doesn’t — is going to become more consequential, not less, as the supply of technically competent output explodes.
There’s something almost liberating about this, if you let yourself sit with it.
A world of commoditized intelligence is, paradoxically, a profoundly human one. It removes the burden of raw computation from the center of what we do and pushes us toward the edges — toward the questions only we can ask, the connections only we can feel, the decisions only we can be held accountable for.
The dealers of intelligence will handle the materials. We’ll still have to decide what to build. Architects.
Questions to Consider
- If intelligence becomes a commodity like electricity or bandwidth, which industries or professions will be slowest to feel that pressure — and why?
- Lessin frames this as a market with many suppliers rather than a winner-take-all race. Does the competitive landscape today support that view, or does it still look like a sprint toward consolidation?
- What does “taste” actually mean when the person exercising it is doing so with AI-augmented perception and judgment? Is it still the same thing?
- Who gets to haggle with the dealers? If cognitive utilities are cheap in aggregate but not universally accessible, does commoditization risk deepening inequality rather than democratizing thought?
- If the value of answering questions falls and the value of asking them rises, what does education need to look like — and how far is it from what it looks like now?