The artificial intelligence (AI) hype seems to bring strong emotions. Some fear automation and think it is going to make everyone jobless and, by extension, poor. Others are enamored with AI and imagine all sorts of wonderful capabilities of these systems.
One of the worst ideas being presented though is this myth that will not die. The idea that a system of production without consumer involvement can be designed. The new iteration is that previous systems were too complex for puny humans. But now AI will continue to develop and will manage to solve all the problems of production.
This is nothing new. It is the old idea of an objective control of production. Socialists believe that an economy can be planned by a group of experts and can “solve” all the problems they believe free markets have. Statists similarly have the outlook of an objective control by either government, the community, or a group of experts to “guide” the economy. And the new breed of statist is claiming that AI will be able to model the economy and take over all production. They even claim this is inevitable: “Jobs will be optional, money meaningless.”
Production
Collectivism and statism in all its forms is based on unsound economic theory. In one of his most important contributions, Mises argued that socialism was impossible because of its inability to perform economic calculation.
And that is the central part of the argument about AI taking over production. How will it know what to produce, and how much of it?
Far from being a planned process, production is a discovery process in which humans are both producers and consumers. Humans act purposely to improve their situation, for example, to satisfy a need to eat. Humans also tend to not be able to produce all they want and need themselves, relying on what others produce. Humans thus will negotiate with others to obtain what others produce. But in exchange they will need to produce what others want.
So production becomes a web of people exchanging with other people, striving to produce what others want to exchange for what they want in return. Ultimately, it is what consumers want that drives the system. With the advent of money and the division of labor, very complex products can be achieved through separate stages in which every individual will do only a small part and exchange it for money.
Value
But it still boils down to what to produce. As there is scarcity on both the means of production and on time, compromises and trade-offs will always need to be made. Even if it is only what to produce first, priorities will need to be made.
This is where value comes in. There have been multiple theories of value—from the cost of production value to the labor theory of value—trying to assign some objective value to goods. There is a tendency to consider that things are valuable in themselves. But this is incorrect. There are multiple goods that one can recognize as being valueless for some and valuable for others—meat for vegans, cigarettes for non-smokers, alcohol for teetotalers.
Menger’s subjective theory of value is the correct one. Humans value goods differently, and that is the basis of exchange. I value a coffee more than the money I have to pay to get it, but the vendor values the money more. (More technically, the vendor values the money because of its purchasing power and ability to exchange it later for other goods). If we valued both things the same, what would be the point of exchange? Both the coffee and the money would satisfy us equally. Exchange would be pointless. In reality exchanges happen because each individual values goods subjectively.
It is only in making or not making an exchange that we can express value for goods. There is no unit of value. I cannot say the coffee gives me three units of value. The way goods are valued is by saying the coffee will satisfy me more at this point than the money. It is by comparison that value can be assigned to goods by individuals.
AI and Socialism
In a production process in which there are no voluntary exchanges then there cannot be any genuine valuation. For AI to produce final consumer goods, AI would have to have all control of the means of production, the same as a central planner. AI would effectively own them. As AI would not be a consumer, it could not exchange anything with another AI, nor with the final consumer.
This is the point Mises made. Without private property and voluntary exchanges in producer goods, economic calculation is impossible. When a government takes revenue in taxes, inflation, or debt, it does not give anything in exchange. This is not a voluntary exchange and thus government cannot value those resources. To government, they are free. If AI controlled all the means of production, the value to consumers would be unclear to the AI too.
When government “provides” a service or gives a subsidy to someone, it is also not through voluntary exchange. When you receive a subsidy, it is “given” to you. In the same way, AI production would need to be “distributed” to humans not through exchange. Those resources would be perceived also as free.
Economic calculation involves valuing what resources you have used to produce something against the value you have obtained from that product. It is the only way to know if it made sense to produce it. If you make a profit, it means you started with something of less value and got something of more value. It made sense to produce. But if you make a loss, it means the resources were more valuable untouched. It made no sense to produce.
That is the way production is made. That is the way we can know what to produce. By seeking the maximum profit, you make what the rest value the most and at the same time you can economize resources as much as possible. In a free market, of course, this is not a smooth operation. Losses happen all the time but they are evident. Each stage calculates if they are making a profit, and, if not, change what they are doing or how they are doing it. This system of interconnected individual exchanges also makes it possible to adapt to changes in what people want, demand, and what resources there are, supply.
Supply and demand are what produce prices and the price system is used for economic calculation, making it possible to coordinate multiple people through space and time to produce complex things such as a pencil.
Impossibility of Economic Calculation
If you cannot assign value to inputs and outputs, you cannot know if it makes sense to produce; you cannot economize. Everything has the same value, everything is free. If supply falls, there is no way to know until the resource disappears. If demand stops, again there is no way to know until unused stock overwhelms the system.
This is the problem with production systems without private property in the means of production and voluntary exchanges. It is not that they are less efficient or give worse outcomes. They are incoherent. This extends to all public endeavors. They only survive in our current economies because they live off real private production and they try to mimic that production.
Thus, governments destroy value by making market valuation impossible or distorted. Once a resource is taken by force, its market value is distorted. AI would do the exact same thing if it took over all the means of production.
Conclusion
In our current world in which socialism is growing alongside increased automation, we need to stress the importance of individuals in the production process. Planned economies are impossible, no matter who is on top.
The automation brought about by AI can be a great tool to increase human productivity and improve everyone’s quality of life as long as we realize how important each person is. The economy is individuals acting to improve our lives and that of others. We should resist anyone trying to stop us from interacting with each other peacefully and trying to control what we do and what we exchange.
