Friedrich von Hayek’s best-known work was the influential and controversial way of easement, published in 1944; but his most celebrated economics writing is The use of knowledge in societya fairly short article on how society obtains and uses scattered information on fundamental economic variables such as preferences, priorities, and productivity.
The article develops a powerful critique of central planning, arguing that centralized authorities are incapable of adequately compiling and processing the “scattered elements of incomplete and often contradictory knowledge held by different individuals.” Because they don’t know what each person prefers among millions of products—let alone their ideas about where to apply their talents most productively and creatively—central planners are doomed to fail.
By contrast, market economies can process and aggregate that information efficiently and effectively: price signals seamlessly convey information about the priorities and preferences of market participants. When tin becomes scarce, its price rises, and, as Hayek explains, “all tin consumers need to know is that some of the tin they used to consume is now being used more profitably elsewhere and that they must therefore economize on its use.”
Nor is this limited to the processing of existing data. The market system, Hayek argues, is also better at discovering—and even producing—relevant new signals: “The data concerning the whole of society from which economic calculation originates is never given to a single mind so that it can deduce its consequences, and never can it be so either.”
Although he is glorified for creating a knowledge-based (“computational”) critique of planning, his arguments are best understood as a call for decentralization in broader terms. He affirms that “if we accept that the economic problem of society is mainly the rapid adaptation to change (…) The final decisions must be left in the hands of those who are familiar with those circumstances.” Ultimately, Hayek concludes, “we have to fix it through some kind of decentralization,” that is, the market economy and the price system.
For decades, Hayek’s arguments served to reject regulations of all kinds. If the regulation of economic activity (such as measures governing the launch of new products) or of prices (such as price caps or price controls) interfere with the functioning of the price system, they will make it difficult for the decentralized process of adaptation to a constantly changing world.
But now artificial intelligence—especially generative AI models that encode, process, and make use of (through hundreds of billions of parameters) gigantic amounts of pre-existing information—presents two challenges to that argument.
First, given AI’s ability to absorb, organize, and interpret data on a massive scale, we might wonder if it would be capable of greater efficiency through central planning than current market systems. That is the hope behind “AI socialism” (or “fully automated fancy communism”): AI will give central planners the means to define optimal and (supposedly) benevolent economic allocations.
But while AI socialism is an interesting thought experiment, it offers only a cursory critique of Hayek. Even if AI were capable of doing all the calculations and data gathering that the market economy already deals with (and that is a very big assumption), the concentration of power in the hands of the central authority would be a major cause for concern.
The famine from which five million Ukrainians died in the early 1930s was not due to Stalin’s inability to calculate proper allocations. On the contrary, he had enough information and used it to get as much grain as possible from the region (due to greater political motivations and perhaps a desire to devastate Ukraine). Furthermore, Hayek’s critique of central planning goes beyond the use of existing data. As we have seen, he focuses primarily on adaptation to change and therefore emphasizes the creation of information as much as its use. “The kind of knowledge to which I have devoted myself,” Hayek writes, “is the kind that by its very nature cannot be part of statistics.” This implies that even an all-powerful large model language (MGL) could not deal with the true nature of sparse information.
But AI presents another, deeper challenge to Hayek’s arguments: in the age of generative AIs like ChatGPT-4, can we even assume that markets will facilitate the decentralized use of information? Alphabet (Google) and Microsoft are leading the development of this technology, they are two gigantic corporations largely dedicated to centralizing information. Even if other companies manage to compete against this duopoly, MGLs, due to their nature, may require a high degree of centralization. It is easy to imagine a scenario in which a large part of humanity obtains its information thanks to the same model.
Of course, Google and Microsoft’s control of information is different from that of the Chinese Communist Party, but as we argue with Simon Johnson in our new book Power and Progress: Our Thousand-Year Struggle over Technology and Prosperity (Power and Progress: A Thousand Years of Trouble in the Face of Technology and Prosperity), even seemingly benign forms of decentralization carry myriad economic and political costs that depend on who is ultimately in control. In the United States, those costs include the increasing monopolization of the technology sector, as data control creates barriers to entry, and the development of business models based on continuous online participation and personalized digital advertisements, which engender emotional outrage, extremism and digital echo chambers, with detrimental effects on democratic participation.
Decentralization is therefore still desirable, but to promote it in the AI age we may need to reverse Hayek’s arguments completely—or at least partly change them—by embracing regulation rather than just focusing on its potential costs.
Follow all the information of Economy and Business in Facebook and Twitteror in our weekly newsletter
Five Days agenda
The most important economic appointments of the day, with the keys and the context to understand their scope.
RECEIVE IT IN YOUR MAIL
Subscribe to continue reading
Read without limits
