Why are we even doing all this?

A guest post of about 1,500 words on Terence Tao’s blog, published on 15 October 2026 by Rikab Gambhir, a theoretical particle physicist. It is satire from first line to last, and it opens with a mock “First they came for…” in which artists, mathematicians and physicists are each offered free tokens and told AI will solve their field.

The occasion is OpenAI’s release in early October of 722 mathematics manuscripts from an internal model. Gambhir calls it every mathematician’s childhood dream: a corporate robot turning out hundreds of obscure, barely readable results as IPO marketing. He sets it beside Borges’s “Library of Babel” (1941), which holds every possible book: every true proof, far more false ones, and, he adds, plenty of “misleading Lean certificates” (linking a paper the same month on why a Lean check of an AI formalisation doesn’t guarantee the proof in English is right). By 2026 standards, he says, the Library would be the most productive research institution in history, and Borges was a fool for never raising venture capital for it.

The joke is that he takes the logic at its word. The breakthrough is generating things that look correct, handing the checking to mathematicians, and dropping understanding as a bottleneck. The supposed consensus online is that research means piling up possibly-true results nobody reads, and explaining them is a waste of resources. He didn’t become a physicist to understand the universe, he says, but to approve histograms from a “Lovecraftian autocomplete”. Joy, learning and beauty are inefficiencies, art is “content”, and he looks forward to ChatGPT enjoying sunsets on his behalf.

The straight point comes at the end. Outsourcing curiosity, creativity and understanding for the sake of productivity changes what it means to be human, and the goal it serves is a complete store of possibly-true facts that nobody reads. He closes by asking readers to remember why they chose science and mathematics.