There’s No “Going Back”: What AI May Erase from Human Experience

Summary

Johan Michalove argues that the spread of generative AI may eventually make pre-AI ways of thinking, working and creating difficult to remember from the inside. He compares this transition with nostalgia for walkable pre-automotive cities and points to AI-generated food illustrations appearing on New York City bodegas as a small example of synthetic media replacing vernacular styles. The essay suggests that future audiences may archive early systems such as diffusion models, GANs and DeepDream as historical markers, even if today’s outputs are often dismissed as disposable “slop.” It also describes unresolved tensions in universities, where students and professors are deciding how much to rely on generative tools, and cites a math problem pursued with large fleets of agents that prompted debate about whether an unintelligible solution fulfills the human purpose of solving difficult problems. In professional communication, the author argues that routing work through a small number of frontier models can produce “Airbrain,” with competing websites, newsletters and advertisements converging on similar language and style. The essay concludes that AI-mediated communication is changing the shared cognitive commons, raising questions about attribution, acceptable machine speech and collective rules. Michalove does not present the transition as predetermined, but calls for clearer language, norms and shared practices that could protect cognitive independence.