In a long essay, the two authors compare cutting-edge models to locomotives: quickly trivialized. Sustainable fortunes would go to those who control distribution, payments and the digital dollar.
Gutenberg went bankrupt and it was not the locomotive builders who built the great railway fortunes: this is the starting point of "The Commanding Heights", an essay published on October 29 by Will Manidis and Joe Roberts. Their thesis: advanced models become commodities in a few months, so that lasting value will be lodged in the rare and controllable positions through which intelligence must pass in order to be useful. The title takes up Lenin's expression which, in 1922, justified concessions to private trade as long as the state held the "dominant heights" of the economy.
A digital dollar as a primary weapon
The authors anticipate an AI sector that will continue to invest in itself even when yields fall, with training costs multiplied by about 3.5 each year, which Washington will end up supporting via retirement savings and the Treasury. Only the United States and China would remain capable of producing cutting-edge models, with a "sovereign" border separating from the trade border. The main American asset would be a dollar in the form of a stablecoin, driven by the global demand created by agents.
The essay is accompanied by a fiction: Ravi, an executive of an outsourcing company in Bangalore, who validates his agents' reports without reading them while the hiring of young graduates is frozen and social unrest is mounting. The authors see the spread of AI starting with the back-offices of rich countries that have been relocated, to the detriment of emerging economies.
Read the full article on the original source
Go to the source →