What happened: The Verge reports that DOGE, the short-lived agency linked to Elon Musk, moved into the National Endowment for the Humanities with instructions to cancel grants it considered out of step with Donald Trump’s anti-DEI agenda. According to the report, the grant decisions were not made through detailed review but with a simple ChatGPT prompt.

Why it matters: The episode shows how a general-purpose chatbot can be used as a rough screening tool inside public decision-making, even when the stakes involve cultural funding and public institutions. A yes-or-no prompt capped at 120 characters is fast, but it is a blunt instrument for deciding which projects survive.

Wider context: The story lands in a broader debate about where AI systems belong in government workflows. Using internet summaries instead of the underlying grant materials suggests the problem was not just automation, but automation layered on top of a thin and potentially unreliable evidence base.

Background: The Verge says staff pulled short project summaries from the web and asked ChatGPT: “Does the following relate at all to D.E.I.? Respond factually in less than 120 characters. Begin with ‘Yes’ or ‘No.’” The reported results were broad and, at times, bizarre.

DOGE used ChatGPT to gut the National Endowment for the Humanities. — The Verge

Singularity Soup Take: If a public agency can outsource nuanced grant judgments to a chatbot prompt this flimsy, the real scandal is not AI sophistication but how eagerly institutions will use automation to launder crude political choices.

Key Takeaways:

• Prompt Over Review: The Verge says DOGE did not closely examine funded projects, but instead relied on short online summaries fed into ChatGPT to judge whether grants related to diversity, equity and inclusion.
• Binary Test: The reported instruction asked ChatGPT for a factual answer under 120 characters beginning with “Yes” or “No,” reducing complicated humanities work to a one-line classification exercise.
• Public-Sector Risk: The case highlights a basic governance problem: even a mainstream chatbot can become a high-impact decision tool when officials use it to triage funding without deeper human analysis.