Smith questions why highly intelligent academics remain relatively low-paid.
In all of human history, has there ever been a commodity with infinite demand, as there appear to be for intelligence? I can't think of one
In economics, rather than modeling the usual price/demand curve to reach an equalibrium demand, perhaps you'd model price/*rate of demand growth*
ChatGPT prefers "recursive expansion of demand" as increasing intelligence opens new applications/markets
Positive users praise recursive or infinite AI-driven intelligence demand as unprecedented and problem-solving, while negative users call the commodity framing fundamentally flawed and criticize economists for enabling societal harm.
No Digg Deeper questions have been answered for this story yet.
In all of human history, has there ever been a commodity with infinite demand, as there appears to be for intelligence? I can't think of one. Even compute, energy or just silicon/sand are just downstream of intelligence, which is the main demand driver.
In economics, rather than modeling the usual price/demand curve to reach an equilibrium, perhaps you'd have to model price/*rate of demand growth* (ie, the derivative of demand, or some other indicator of velocity)
Interestingly, ChatGPT (below) prefers the framework of "recursive expansion of demand" as increasing intelligence opens new applications/markets.
But the end result is the same -- the demand curve keeps moving to the right, maybe forever.
Which I think is unprecedented.
A lot of answers to this question, but zero satisfying ones so far.
For those pointing to moments of steep commodity demand growth in the past, let me make my point clearer about what's different about intelligence.
it's not just that demand for intelligence is parabolic at the moment or that it's happening faster than commodity growths in the past, it's that unlike previous commodities, I can't see any theoretical limit to it.
All previous commodities were gated by human demand. But intelligence is the first commodity that *creates its own demand* and is not limited by the carrying capacity of the planet.
In all of human history, has there ever been a commodity with infinite demand, as there appears to be for intelligence? I can't think of one. Even compute, energy or just silicon/sand are just downstream of intelligence, which is the main demand driver.
In economics, rather than modeling the usual price/demand curve to reach an equilibrium, perhaps you'd have to model price/*rate of demand growth* (ie, the derivative of demand, or some other indicator of velocity)
Interestingly, ChatGPT (below) prefers the framework of "recursive expansion of demand" as increasing intelligence opens new applications/markets.
But the end result is the same -- the demand curve keeps moving to the right, maybe forever.
Which I think is unprecedented.
@chr1sa @harris_edouard yes, TAM is the really big moving target here
imo, the biggest uncertainty in the AI supercycle is the actual TAM of frontier LLMs. people have mental models between "consumer chat app", "useful API", "coding assistant", "high disruption white collar productivity tool", "automate all labour", "the future" and "god".
$NVDA
Indeed, AI is driving a continuous expansion of both intelligence and the demand for knowledge.
How can we make AI agents truly trustworthy?
To achieve real results, simply replacing humans with technology is not enough. We must focus on preserving jobs while learning how humans and AI can collaborate as a true team. Once we unlock this synergy, the demand for AI will skyrocket.
But how exactly will this mass adoption happen?
To use an aviation analogy, current AI is still at the "Concorde" stage—highly advanced, yet limited in reach. It has not yet achieved the mass-scale "democratization of demand" that the DC-9 brought to air travel.
To truly democratize AI and spread its benefits widely, we must move beyond using it as a mere tool for cost-cutting. What AI agents need now is to provide the same level of safety, reliability, and accessibility that the DC-9 once did.
@johnway @paulg No. You can only eat, breath or drink so much. But intelligence doesn't even need you to be useful. Human limits do not apply
Interesting question. I will guess the demand is better reflected by funding than by the supply of academics. I think demand for intelligence will indeed have no upper limit, but I don’t think it is ever infinite. If intelligence creates demand for more intelligence there will still be a physical timescale that determines the exponential growth rate.
@Noahpinion Because they’re not the smartest people.
The best measure of intelligence is likely wealth, not IQ. Generating wealth requires far more intelligence than getting a PhD in a very narrow field.
A billionaire high school dropout is likely far smarter than a Nobel prize winner.
For funsies, read Accelerando. In that scenario marginal costs per compute unit fall as rising intelligence makes it cheaper to harness all the universe's energy and matter. Dyson spheres, computronium, etc.
Not saying it's likely or even practical, but it's helpful to think beyond Earth/humans.
@Noahpinion Just creating conditions for proper open-ended research (the kind OpenAI has just started to emulate) and selecting for people willing to do that.
I've wondered about this. If demand for intelligence were so infinite, why do we have so.many of our smartest people sitting around doing esoteric fun stuff for smallish salaries in academia?
@Noahpinion Humans have agency. There are intelligent people who are “irrational” actors under capitalism who are not efficiently sorted into neat little jobs that generate the most value. Sometimes people like doing other things, and capital is not their priority.
Wealth in America is not a neutral IQ test. It is a measure of who gets access to ownership, leverage, inheritance, networks, monopoly protection, cheap capital, tax shelters, and state-backed asset inflation.
A billionaire high-school dropout is not automatically smarter than a Nobel Prize winner. He may simply be closer to capital, better positioned inside a monopoly, born into a network, sitting on equity, or operating inside a system where ownership compounds faster than labor, research, or discovery.
A PhD physicist may understand the structure of the universe. A billionaire may own equity in a platform with network effects, regulatory capture, buybacks, tax deferral, and monopoly rents.
Those are not the same thing.
Wealth measures control over claims on future cash flows.
It does not measure intelligence in any clean moral or scientific sense.
To admit that is to puncture one of America’s central myths: meritocracy.
Because if wealth is not proof of superior intelligence, then billionaire power is not moral destiny. It is structure. It is access. It is ownership. It is law. It is inheritance. It is proximity to capital dressed up as genius.
In America, wealth is not an IQ score. It is a receipt for proximity to capital.
@alanrosca I find this an unconvincing answer. I don't think simply choosing different definitions of "intelligence" offers much insight in this case.
@chr1sa The supply curve for querying AI at some point slopes upward, because of energy intensity and the need to replace/upgrade the chips.
Rising marginal cost will limit usage, as always.
Hmm. What I said is probably a sloppy formulation, but I think there’s something real happening in all the exponential growth metrics, and the obvious explanation is that they are the solution to the differential equation:
dY/dt = alpha*Y
namely
Y = C exp(alpha*t)
where Y is **something** that drives the process — is it technology, information, intelligence,…?
A few years ago I reviewed the literature on proposed explanations for Moore’s Law and similar tech metrics. The only good paper I found said that this “Y” is the overall development of the technological economy. Moore’s Law is seen as simply one of many measurable indicators of that overall development and they all go along for the ride showing exponential growth. I think that makes sense.
But that doesn’t explain what underlies the demand in the first place. My thinking is that intelligence is an aspect of Y though it’s probably more complicated than that.
@Noahpinion I make this point frequently. There are good reasons to expect strongly diminishing returns to intelligence. An example is most image classification. It is overkill to use frontier models. Alternatively, context or coherence rather than intelligence may be the bottleneck.
@Noahpinion perhaps lots of reasons .. they are often over rated intelligence .. over valued, they might also know it (motivation not being found out ) .. they don't have utility needs average intelligence do .. (motivation for money lessened ) .. mechanized intelligence is different
@chr1sa I think most analogous commodity would be energy, for which I'd argue there is infinite demand at the right price. As the price comes down, new use cases get unlocked driving new sources of demand.
The price demand curve doesn’t break down because intelligence doesn’t have infinite demand.
Everyone needs as much intelligence as they can get right?
But the same is true of water or food or oxygen. The demand isn’t infinite, the demand per capita is just really really high. People have diminishing returns.
So when looking at it on a global scale, the demand for intelligence — demand per person times number of people + AI instances — is insanely high (and literally has always been) but is not infinite.
@Noahpinion As the bar to be gainfully employed keeps increasing, i.e. the ‘smarts’ floor to even have a job keeps rising, we’re going to keep seeing academia poached for nothing but access to sheer intelligence