I suggest that Jevon's Paradox applies to the product of AI. So what is that product? I maintain that it is words: not meaningful or useful content, just verbiage. Indeed, it's arguable that the failure to realise the productivity gains expected over the past couple of decades is due to the increasing demand for verbiage: that is, a significant proportion of the white collar class is employed in producing and consuming verbiage with little to no effect whatsoever on the production of economic goods. This relates to David Graber's Bullshit Jobs hypothesis. I suggest that verbiage, not knowledge, is what many "knowledge workers" produce.
As an example: the Lower Thames Crossing has required 2,000 documents, 360,000 pages, about 100 million words: costing £267 million to produce before permission could be obtained to start work. What will happen when the cost of producing and consuming such documents falls by a factor of a million? Will we see the cost of producing the planning documents for the next such project fall to £267? Or will the total amount of verbiage produced rise to 100 trillion words? My guess is that the latter will be closer to the truth.
For Jevon’s paradox to apply to the use of intelligence of humans for humans, there must be a limit to how much AI intelligence can replace human intelligence applied for human goals. Otherwise, even if there are multiplying uses for intelligence as increasing amounts become available, those uses may be for humans or for AI, and either way, rely on increasing use of AI intelligence, not human intelligence.
I agree with the message of this, but I think the reason that the labor market will not be annihilated in the medium to long-term is because the types of jobs that people are gonna be doing will change. I do think many roles that we see in the white collar space will likely disappear.
I think that's consistent with the argument here! I guess I'm skeptical they'll disappear fast enough to cause much of a short-term bump in unemployment. Seems more likely to me that existing roles will be gradually redefined, rather than being cut and replaced.
I don’t know the timeline—my confidence bands are too wide to say something useful w/onmore data. But the details matter here for welfare: 10 years of unemployment before re calibrating is much worse than gradual trend.
For sure. Maybe this reads too much into a small amount of evidence, but I find the radiology and translation examples compelling. Seems like AI's been strong enough in both fields for long enough that by now you'd expect some greater signs of disruption.
> enable more granular and accurate financial forecasting, offer more expansive
> scenario-modeling capabilities, and help firms provide more insightful tax strategies
> and decision-making support based on real-time analysis of much larger sets of data
I hear this "humans will just move to higher order tasks" all the time, and it is clearly just cope. AI is already better than humans at all of these things, and will only get better from here.
Jevons says when you get more efficient at using a resource, you discover new uses for it, and total consumption goes up. Coal gets cheaper to burn, you find more things to do with coal, you burn more coal.
But the resource that matters is not human labor. The resource is intelligence, and AI is a new, cheaper source of it. Yes, cheaper intelligence means we'll find tons of new uses for it. But those new uses will largely be fulfilled by AI, not by hiring more people. We'll have the same or fewer humans producing dramatically more output.
It's increasingly feeling like this debate boils down to whether future AIs will be *perfect* substitutes for all intellectual human labor, or whether they will fall short of that, however slightly.
As long as there is some complementarity left, human productivity will explode and new jobs will start blooming at different sectors of the economy.
However, I think it's worth taking seriously the prospects of actually, genuinely, human-level AI. And the possibility that all those final touches that currently need a human designer, all the prompt engineering that's making devs more productive, all of that plus the new jobs that were created during the complementarity phase, at some point, could be automated by an advanced AI.
No one knows whether the current LLM paradigm will take us there or exactly what breakthroughs are going to be needed (though there are some promising ideas, like continual learning, embodiment etc,). Nevertheless, relying on that "human spark" to magically ensure that we stay relevant, that we stay "in the loop", no matter how advanced AIs might get, reads a bit like an argument from lack of imagination.
There are many examples in which demand elasticity was exceeded and employment collapsed, agriculture being the big one (about 10 fold employment contraction in the US, considering also related sectors, not just farm work). If the hype about AI is at least partially correct we are going for a big socio-economic experiment and the honest thing to say is that we don't know what will come out on the other side. Will humans find a way to be complementary to computers? Will completely new sectors emerge? If AGI becomes real, think for a moment, what do you have to offer to an employer? The "G" in AGI, taken in its strictest sense, means the answer is "nothing". There will be nothing you can do that a machine can't. Your salary will have to go down to the cost of an LLM for the same task. I for one hope the hype is excessive, because the last think mankind needs presently is a vast increase in energy needs and massive unemployment.
Every object will become enchanted. We are going towards a world of animism. Intelligence will be imbued into everything, to the degree we find it desirable. Our cars will be as intelligent as horses, and probably a bit more. But they won't be as intelligent as e.g. Louis Borges. We have no need for that. Intelligence and life are almost synonymous to us at our present experience. ("intelligence is the computational part of achieving a goal"; life goal#1 is staying alive) With intelligence imbued in our everyday objects, they will come 'alive' to a degree. It's extraordinary ngl: the most abundant material on Earth - sand, silicate - to be made implement the rarest of things in our universe - intelligence.
I suggest that Jevon's Paradox applies to the product of AI. So what is that product? I maintain that it is words: not meaningful or useful content, just verbiage. Indeed, it's arguable that the failure to realise the productivity gains expected over the past couple of decades is due to the increasing demand for verbiage: that is, a significant proportion of the white collar class is employed in producing and consuming verbiage with little to no effect whatsoever on the production of economic goods. This relates to David Graber's Bullshit Jobs hypothesis. I suggest that verbiage, not knowledge, is what many "knowledge workers" produce.
As an example: the Lower Thames Crossing has required 2,000 documents, 360,000 pages, about 100 million words: costing £267 million to produce before permission could be obtained to start work. What will happen when the cost of producing and consuming such documents falls by a factor of a million? Will we see the cost of producing the planning documents for the next such project fall to £267? Or will the total amount of verbiage produced rise to 100 trillion words? My guess is that the latter will be closer to the truth.
Thank you for this positive vision!
For Jevon’s paradox to apply to the use of intelligence of humans for humans, there must be a limit to how much AI intelligence can replace human intelligence applied for human goals. Otherwise, even if there are multiplying uses for intelligence as increasing amounts become available, those uses may be for humans or for AI, and either way, rely on increasing use of AI intelligence, not human intelligence.
Re graduate unemployment, this overview article of mine might be of interest: https://futureecon.substack.com/p/ai-is-breaking-the-career-ladder
Thanks for this! Will give it a close read when I have a moment free.
I agree with the message of this, but I think the reason that the labor market will not be annihilated in the medium to long-term is because the types of jobs that people are gonna be doing will change. I do think many roles that we see in the white collar space will likely disappear.
I think that's consistent with the argument here! I guess I'm skeptical they'll disappear fast enough to cause much of a short-term bump in unemployment. Seems more likely to me that existing roles will be gradually redefined, rather than being cut and replaced.
I don’t know the timeline—my confidence bands are too wide to say something useful w/onmore data. But the details matter here for welfare: 10 years of unemployment before re calibrating is much worse than gradual trend.
For sure. Maybe this reads too much into a small amount of evidence, but I find the radiology and translation examples compelling. Seems like AI's been strong enough in both fields for long enough that by now you'd expect some greater signs of disruption.
I’ll have a post on this within a week, working out the math for the how and the why. Let’s see if my argument is compelling.
Looking forward to it!
> enable more granular and accurate financial forecasting, offer more expansive
> scenario-modeling capabilities, and help firms provide more insightful tax strategies
> and decision-making support based on real-time analysis of much larger sets of data
I hear this "humans will just move to higher order tasks" all the time, and it is clearly just cope. AI is already better than humans at all of these things, and will only get better from here.
Jevons says when you get more efficient at using a resource, you discover new uses for it, and total consumption goes up. Coal gets cheaper to burn, you find more things to do with coal, you burn more coal.
But the resource that matters is not human labor. The resource is intelligence, and AI is a new, cheaper source of it. Yes, cheaper intelligence means we'll find tons of new uses for it. But those new uses will largely be fulfilled by AI, not by hiring more people. We'll have the same or fewer humans producing dramatically more output.
I'm very worried about what's ahead.
It's increasingly feeling like this debate boils down to whether future AIs will be *perfect* substitutes for all intellectual human labor, or whether they will fall short of that, however slightly.
As long as there is some complementarity left, human productivity will explode and new jobs will start blooming at different sectors of the economy.
However, I think it's worth taking seriously the prospects of actually, genuinely, human-level AI. And the possibility that all those final touches that currently need a human designer, all the prompt engineering that's making devs more productive, all of that plus the new jobs that were created during the complementarity phase, at some point, could be automated by an advanced AI.
No one knows whether the current LLM paradigm will take us there or exactly what breakthroughs are going to be needed (though there are some promising ideas, like continual learning, embodiment etc,). Nevertheless, relying on that "human spark" to magically ensure that we stay relevant, that we stay "in the loop", no matter how advanced AIs might get, reads a bit like an argument from lack of imagination.
There are many examples in which demand elasticity was exceeded and employment collapsed, agriculture being the big one (about 10 fold employment contraction in the US, considering also related sectors, not just farm work). If the hype about AI is at least partially correct we are going for a big socio-economic experiment and the honest thing to say is that we don't know what will come out on the other side. Will humans find a way to be complementary to computers? Will completely new sectors emerge? If AGI becomes real, think for a moment, what do you have to offer to an employer? The "G" in AGI, taken in its strictest sense, means the answer is "nothing". There will be nothing you can do that a machine can't. Your salary will have to go down to the cost of an LLM for the same task. I for one hope the hype is excessive, because the last think mankind needs presently is a vast increase in energy needs and massive unemployment.
Every object will become enchanted. We are going towards a world of animism. Intelligence will be imbued into everything, to the degree we find it desirable. Our cars will be as intelligent as horses, and probably a bit more. But they won't be as intelligent as e.g. Louis Borges. We have no need for that. Intelligence and life are almost synonymous to us at our present experience. ("intelligence is the computational part of achieving a goal"; life goal#1 is staying alive) With intelligence imbued in our everyday objects, they will come 'alive' to a degree. It's extraordinary ngl: the most abundant material on Earth - sand, silicate - to be made implement the rarest of things in our universe - intelligence.