The Atlantic's diagnosis is clean: AI has already changed writing. Now it is changing what it means to read. The anxiety is real, but it is also slightly mislocated. The thing AI changes is not reading itself. It is the reader's ability to trust that something was written. Which is a different problem entirely, and an older one than it seems.
Authenticity, Attribution, and the Archive
Consider what surfaced this week in parallel. Newly unearthed John Lennon drawings, made with artist Stephen Verona in 1964, are making their public debut. A newly confirmed Lucian Freud painting, whose attribution concluded a decades-long struggle, debuts in London. And Emily Sargent watercolors hidden in a trunk arrive at auction. Three stories about works whose value depends entirely on the certainty of human origin. The market for authenticated objects is not a coincidence. It is a hedge against the exact condition The Atlantic is describing.
The Restlessness Is Structural, Not Psychological
The arXiv paper most relevant here is not one about art. It is a 2026 paper by de Souza et al. on PEEL as semiotic scaffolding for AI-enabled research, which argues that LLMs are quietly eroding researchers' epistemic accountability, their ability to own the reasoning chain behind a claim. This is the reading problem The Atlantic is circling. When you cannot trust the provenance of a text, the act of reading becomes forensic. You are no longer absorbing an argument. You are auditing one. The Lennon drawings and the Freud painting are the inverse: objects whose provenance was already forensically established, which is precisely what makes them valuable now. TurboFund's piece on flipping screen time from passive to active touches this shift: the platforms and products winning right now are the ones that turn consumption into participation, because passive reading is the mode AI has most thoroughly colonized. Restless reading is not a failure of attention. It is a rational response to an environment where the author may not exist.