DOGE let ChatGPT do the thinking

A friend once complained about the fruitlessness of raising her son not to play with guns. He didn’t get them as toys. she said. It didn’t matter. Everything became a gun: a stick is a gun, a banana is a gun, etc. It’s like that with Republicans. Points for imagination. Even the bluntest object becomes a weapon against Republicans’ adversaries: “death panels,” Christmas, “the great replacement,” critical race theory, “woke,” clean energy, etc. Or DEI (diversity, equity, and inclusion).
It’s not just that Republicans latch onto these as rhetorical bludgeons. Give them the opportunity and they become legal bludgeons that do real harm. For example, how Donald Trump, Elon Musk, and his DOGE goons, in coordination with a supine Republican congressional delegation — and with the help of ChatGPT — set out to rid the United States of all things DEI.
The American Historical Association and others filed suit over the capricious and illegal cancelling of grant programs last year. They provided an update last week after plaintiffs filed for summary judgment in the case. Among what came out in discovery was the fact that National Endowment for the Humanities (NEH) acting chair Michael McDonald ceded his authority over grants to Musk’s Doge programming geeks. The DOGE kids let ChatGPT do their thinking for them:
- McDonald and key members of the DOGE team bypassed authorized record preservation requirements and violated the Federal Records Act by conducting official government business regarding the cuts using Signal, a messaging application unauthorized for federal employees, and intentionally set to automatically delete messages.
- DOGE fed grant descriptions into OpenAI’s ChatGPT generative artificial intelligence chatbot, asking it to decide if grants were “DEI.” They then entered ChatGPT’s responses into a spreadsheet compiling all NEH grants, including its “DEI rationale” and “Yes / No DEI?” replies. This ChatGPT-generated list was used in place of the list created by NEH staffers to identify which grants to cut. Projects Grants that were flagged as “DEI” and then terminated included a documentary sharing the story of Jewish women’s slave labor during the Holocaust; an archival project on the lives of Italian Americans; a project to digitize photograph collections of Appalachian residents; and multiple projects to preserve endangered Native American languages and cultures.
- DOGE staffers violated the Federal Equal Protection Clause of the 5th Amendment by flagging grant descriptions as “DEI” solely because they included “BIPOC (Black, Indigenous, People of Color),” “homosexual,” “LGBTQ,” and “Tribal,” among other terms.
- DOGE staffers also flagged grants that NEH leaders concede had no connection to DEI, including grants that had been awarded for collections management after a natural disaster, preservation training, and improving HVAC systems.
Asked in deposition how DOGE went about its work, DOGE staffer Justin Fox explained as if caught in a programming loop. Behold your DOGE overlords:
The entire 28-minute Fox deposition is here.
The Ink picks out another exchange between Authors Guild/AHA counsel Yinka Onayemi and Fox that is just as weird:
Onayemi (reads from the DOGE spreadsheet of grants recommended for termination by ChatGPT): “The documentary tells the story of the Colfax massacre, the single greatest incident of anti-Black violence during Reconstruction, and its historical legacy for black civil rights in Louisiana, the South, and the nation as a whole. Did I read that correctly?
Fox: “Yes.”
Onayemi: “Okay. And then in column B, right next to that, it says ‘The documentary explores a historical event that significantly impacted Black civil rights, making it relevant to the topic of DEI.” Did I read that correctly?
Fox: “Yes.”
Onayemi: “Is it fair to say that what I just read is the ChatGPT output of the prompt in the first column?
Fox: “Yes.”
Onayemi: Do you agree with ChatGPT’s assessment here, that the documentary is DEI if it explores historical events that heavily impacted Black civil rights?
Fox: “Yes.”
Onayemi: “Why would that be DEI?”
Fox: “It’s focused on a singular race. It is not for the benefit of humankind, it is focused on a specific group or a specific race, here being Black.”
Onayemi: Why would learning about anti-Black violence not be to the benefit of humankind?
Fox’s counsel (cutting off this line of questioning): “Objection.”
This isn’t a matter of a keyword-search snafu or poor A.I. training — though plenty of that comes up in the conversations with Fox and his colleague Nathan Cavanaugh — things like the cancellation of an upgrade to an HVAC system at a North Carolina museum because of a mission statement that includes serving “diverse audiences” — and those are bad enough.
What these exchanges reveal, The Ink suggests, is that “DOGE’s vision of humankind (at least the part of it that’s worthy of benefiting from public investment) is necessarily white and male.”
And that’s the vision being encoded into the tools of the future, being charged with the education of future generations. That’s a cautionary tale, unfolding in real time, as widespread anxiety boils over about the degree to which America is turning over higher education to A.I. — often in partnership with OpenAI, and in tandem with the Trump administration’s plan to remake universities in its own image. And the fact that academics are fighting back, with historical understanding, is about the best case for the power of actual, tangible, critical historical understanding — the kind of thing the humanities exist to instill in human beings and their institutions.
But it’s more than just A.I. developers encoding it as a tool of white supremacy. Not only have DOGE idiots allowed A.I. to strip government oversight from Congress, they’ve clearly settled their brains into WALL·E -style hover chairs and let A.I. do their thinking for them. Allegedly, they were hired for being boy geniuses. Except they didn’t question ChatGPT’s output. It came out of a computer, right?
That’s been the bane of my struggles with Democrats to think outside the VoteBuilder box (their national targeting tool) and begin seeing the voter forest, not just the trees. What the programming gods coded into the algorithm must be right, right? (If average users even think that hard about it.) And what the computer spits out must be right.
I was a consulting engineer for 35 years. I knew better than to let the tool do my thinking. Half the time I log into the Democrats’ database, I hear in my head a very familiar voice issuing this classic warning.
Republicans, Trump, Musk, and DOGE (and Democrats) put too much faith in the technological terrors they’ve constructed. All of us suffer for it.








