For most people, artificial intelligence remains an abstract technology. In his newsletter Interconnects, the author argues that comparisons between the current AI boom and past industrial revolutions miss a crucial factor: how change concretely affects daily life. According to him, tangible benefits remain rare, and society offers more inertia than in previous eras.
Returning after a few weeks away for his wedding in New England, he observes that the general public’s contact with AI products is marginal, not very useful, or even confusing. Many people have heard about the OpenAI-HuggingFace incident without knowing what to think. On the positive side, AI is mainly used to create fun images or enhance Google search. On the negative side, it is associated with addictive social media algorithms, relatives hooked on chatbots, and a burgeoning debate about data centers.
Family, food, transportation, or leisure are, according to the author, still little affected. He recalls that industrial revolutions had transformed ordinary life. The First Industrial Revolution, in the late 18th century, made cheaper clothing and kitchen utensils, books, and new jobs accessible. The Second, in the late 19th century, introduced domestic appliances, preserved foods, running water in homes, photography, or the bicycle. These very concrete advances are still used by many today.
Indirect Benefits and Slow Diffusion
The most ambitious promises of AI, such as new scientific discoveries, treatments for rare diseases, or lasting economic abundance, risk being too indirect, he believes. “How will an ordinary citizen come to credit OpenAI or Anthropic with saving their life, if they first consulted their family doctor, who told them about a new miracle cure?” he asks. He also wonders how many Americans will care about OpenAI solving the Navier-Stokes Millennium Prize problem.
For the author, the current phase serves above all to build infrastructure and processes that will accumulate over decades. A major mathematical breakthrough will likely seem modest compared to subsequent progress. He judges it probable that, in fifty years, the average American’s daily life will closely resemble today’s, in terms of housing, appliances, relationships, and vehicles—excluding autonomous driving, which follows a trajectory distinct from large language models. AI would then receive a large share of the credit.
An Imbalance Creating Tensions
This gap worries the author. Intellectual work represents about half of the U.S. economy, and AI would be “as fundamental as electricity,” or will become so quickly thanks to expected progress in agents—systems capable of autonomously chaining tasks—in the next 18 months. He judges current AI mainly useful to an elite, while the tech economy prospers as daily life stagnates for many.
Looking at the Engels’ pause, he cites the definition of this period: “the period from 1790 to 1840, during which the wages of the British working class stagnated and GDP per capita grew rapidly during a technological upheaval.” If this parallel were the closest to what awaits AI, those who do not benefit from it would be right to protest. He also does not expect the tech industry to increase its headcount, agreeing on this point with Doug OLaughlin, who suggests headcount could shrink while intellectual work output explodes.
He fears such a reaction could hinder AI development, steering it toward the scenario of American nuclear energy. According to him, young people following this topic today will see AI go from almost 0% to over 90% complete adoption in their lifetime.
