Disclaimer 1: So far, I raised this point in a few conversations, and people encouraged me to develop my thinking into a post. This is my attempt to do so. I have no expert knowledge of economics.
Disclaimer 2: What I call Transformative AI (TAI) here is something that will automate existing jobs but will remain under human control. This is not the typical definition I encountered, which is AI that will cause GDP growth per year of the order of 20% − 100%. For the purpose of thinking-in-limits, the difference is not essential.
preambular
One of the things they teach you in physics is to think in limits. This means seeing how the solution will look like when the variable is put into its extreme value, often infinity or zero. Surprisingly enough, in some cases, the proper solution is the product of the solutions of two simple solutions that pop up in the opposite limits of the variable. TAI can be seen as the limit of production efficiency. Here are my thoughts on what can happen to economics exploring the limits of consumption as an essential counterpart of production.
1. Demand saturation
There are two types of businesses: those that meet existing needs and those which create new needs and then meet them. The latter, which would include Facebook, Uber, or Airbnb, is a rather rare although noticeable case
This distinction is important for the following fundamental question: ”Is there a saturation of human needs?” or, which is the same, “Is there demand saturation?”.
Clearly, with improved efficiency, we will be able to saturate the existing needs—automate the production away, e voila. With ever more efficient production tools, there will be a transition period of scaling those tools, but in the limit, the production will be contained only by finite resources. More on this in the section below.
Now, given the existing needs will be saturated (you can only eat so much food, take so many trips, and watch this much Netflix), the question arises: Will we be able to create new needs continuously? The limits here are: 1. Consumption time. Indeed, it’s probably not worth buying a yacht if you have no time to sail it. 2. Purchase capacity. Indeed, if I’m out of a job, I might not be able to pay for the newest, most amazing service. I would suspect this is a limit that we are currently hitting in our economy. One solution that is currently heavily employed is debt. More on this in Section 3
The actual “final state” will be defined by the limit we hit first. If my guess on the purchase capacity limit is correct, TAI will only accelerate the existing trend of increasing inequality. If no radical economic restructuring is introduced (e.g., universal basic income), and the TAI capital is conscious of not allowing mass poverty and thus civil unrest to set into place, the TAI economy will look very much like the modern economy—having tonnes of workforce who produce services of questionable value just to keep the population engaged in the economy. This is a very critical view, however, terms like Bullshit_Jobs and tiktok economy display some features of the modern economy that probably would have been considered dystopian in the mid-XX century. This is still “okay,” though, for I doubt the reader of this post is a blue-collar worker and, as such, is already a proud citizen of our common simulacra. Keep calm and sell signs services!
2. Resource wars
As mentioned above, resources are limited. Imagine we are at the point where robots learned how to construct factories to build more compute, more cars, more buildings, more robots, and more power plants (solar, nuclear, fusion). And imagine that those robots are cheap. Well, probably before we hit the consumption limit, we will hit the resource limit, at least for uranium and rare earth minerals. There is a reason the word “rare” is in the name. Now, this is when it will be decided who gets to be rich and who gets to be the rest. Control the resources, defend them enough, and you are winning the race. And, of course, the resources define not only the regular economy but also the defense capacity.
Today, the US and China have substantial reserves of rare earth elements. GPT4 estimates US reserves to last for 250 years, which means that this limit will be hit if the production of electronic equipment scales up a factor of ~100.
3. Debt
Debt is a solution to consumption capacity exhaustion. A peculiar norm existing now in Switzerland is the never-pay-off mortgage. You see, real estate is so expensive here that many people cannot afford to pay it off in their lifetime. Well, when you die, your kids can decide if they want to keep paying. Many people think it’s a great deal since the interests are lower than the rate. Yet, if one accounts for a 20% down payment, it is basically the same. An extra beautiful passage is that you can use your pension for the down payment. Buy now, think later.
Apart from real estate and your pension, what else can you provide as a security for the debt? Dystopian scenarios would include some forms of freedom limitation (e.g., you’d agree to relocate if needed, you’d sign some NDAs that limit your freedom of speech, or you’d sell your voting rights). As bad as it sounds, these already exist in some rudimentary forms nowadays in poor societies. My relative used to sell her voice for a monthly supply of sunflower oil and sugar during turbulent years of young Ukrainian democracy.
Where I am wrong
Thinking-in-limits does not always work in physics. It assumes that the problem is still a well-formulated problem, even in the limit. This is not always the case and probably is not the case in economics. Most likely, we’ll go through a series of “phase transitions.” That is, new qualitative change will emerge that will not obey modern market economy rules.
An optimistic (fantastic) subjective view of possible emergent phenomena: - universal basic income - abundance kills consumerism - economy reshapes into targeting spiritual values
Of course, there’s the whole zoo of other dystopian scenarios I don’t describe.
Why this is useful
Assuming we are on the trajectory to the limits, the described effects can still be counteracted with policy. Policy is slow, so it’s better to think about it sooner than later.
What will I do about it
- educate myself on economics, conflict, and forecasting - have a plan A B to not to get poor too soon - see the reaction to this post (if any) → update, if confirmative—spread the word.
Thinking-in-limits about TAI from the demand perspective. Demand saturation, resource wars, new debt.
Epistemic status: considerations
Disclaimer 1: So far, I raised this point in a few conversations, and people encouraged me to develop my thinking into a post. This is my attempt to do so. I have no expert knowledge of economics.
Disclaimer 2: What I call Transformative AI (TAI) here is something that will automate existing jobs but will remain under human control. This is not the typical definition I encountered, which is AI that will cause GDP growth per year of the order of 20% − 100%. For the purpose of thinking-in-limits, the difference is not essential.
preambular
One of the things they teach you in physics is to think in limits. This means seeing how the solution will look like when the variable is put into its extreme value, often infinity or zero. Surprisingly enough, in some cases, the proper solution is the product of the solutions of two simple solutions that pop up in the opposite limits of the variable.
TAI can be seen as the limit of production efficiency. Here are my thoughts on what can happen to economics exploring the limits of consumption as an essential counterpart of production.
1. Demand saturation
There are two types of businesses: those that meet existing needs and those which create new needs and then meet them. The latter, which would include Facebook, Uber, or Airbnb, is a rather rare although noticeable case
This distinction is important for the following fundamental question:
”Is there a saturation of human needs?” or, which is the same, “Is there demand saturation?”.
Clearly, with improved efficiency, we will be able to saturate the existing needs—automate the production away, e voila. With ever more efficient production tools, there will be a transition period of scaling those tools, but in the limit, the production will be contained only by finite resources. More on this in the section below.
Now, given the existing needs will be saturated (you can only eat so much food, take so many trips, and watch this much Netflix), the question arises: Will we be able to create new needs continuously? The limits here are:
1. Consumption time. Indeed, it’s probably not worth buying a yacht if you have no time to sail it.
2. Purchase capacity. Indeed, if I’m out of a job, I might not be able to pay for the newest, most amazing service. I would suspect this is a limit that we are currently hitting in our economy. One solution that is currently heavily employed is debt. More on this in Section 3
The actual “final state” will be defined by the limit we hit first. If my guess on the purchase capacity limit is correct, TAI will only accelerate the existing trend of increasing inequality. If no radical economic restructuring is introduced (e.g., universal basic income), and the TAI capital is conscious of not allowing mass poverty and thus civil unrest to set into place, the TAI economy will look very much like the modern economy—having tonnes of workforce who produce services of questionable value just to keep the population engaged in the economy. This is a very critical view, however, terms like Bullshit_Jobs and tiktok economy display some features of the modern economy that probably would have been considered dystopian in the mid-XX century. This is still “okay,” though, for I doubt the reader of this post is a blue-collar worker and, as such, is already a proud citizen of our common simulacra. Keep calm and sell
signsservices!2. Resource wars
As mentioned above, resources are limited. Imagine we are at the point where robots learned how to construct factories to build more compute, more cars, more buildings, more robots, and more power plants (solar, nuclear, fusion). And imagine that those robots are cheap. Well, probably before we hit the consumption limit, we will hit the resource limit, at least for uranium and rare earth minerals. There is a reason the word “rare” is in the name. Now, this is when it will be decided who gets to be rich and who gets to be the rest. Control the resources, defend them enough, and you are winning the race. And, of course, the resources define not only the regular economy but also the defense capacity.
Today, the US and China have substantial reserves of rare earth elements. GPT4 estimates US reserves to last for 250 years, which means that this limit will be hit if the production of electronic equipment scales up a factor of ~100.
3. Debt
Debt is a solution to consumption capacity exhaustion. A peculiar norm existing now in Switzerland is the never-pay-off mortgage. You see, real estate is so expensive here that many people cannot afford to pay it off in their lifetime. Well, when you die, your kids can decide if they want to keep paying. Many people think it’s a great deal since the interests are lower than the rate. Yet, if one accounts for a 20% down payment, it is basically the same. An extra beautiful passage is that you can use your pension for the down payment. Buy now, think later.
Apart from real estate and your pension, what else can you provide as a security for the debt? Dystopian scenarios would include some forms of freedom limitation (e.g., you’d agree to relocate if needed, you’d sign some NDAs that limit your freedom of speech, or you’d sell your voting rights). As bad as it sounds, these already exist in some rudimentary forms nowadays in poor societies. My relative used to sell her voice for a monthly supply of sunflower oil and sugar during turbulent years of young Ukrainian democracy.
Where I am wrong
Thinking-in-limits does not always work in physics. It assumes that the problem is still a well-formulated problem, even in the limit. This is not always the case and probably is not the case in economics. Most likely, we’ll go through a series of “phase transitions.” That is, new qualitative change will emerge that will not obey modern market economy rules.
An optimistic (fantastic) subjective view of possible emergent phenomena:
- universal basic income
- abundance kills consumerism
- economy reshapes into targeting spiritual values
Of course, there’s the whole zoo of other dystopian scenarios I don’t describe.
Why this is useful
Assuming we are on the trajectory to the limits, the described effects can still be counteracted with policy. Policy is slow, so it’s better to think about it sooner than later.
What will I do about it
- educate myself on economics, conflict, and forecasting
- have a plan
AB to not to get poor too soon- see the reaction to this post (if any) → update, if confirmative—spread the word.