Tech elites are turning AI into ChatGPTrump
The administration's new AI plan gives Silicon Valley everything it wants — for a price.
🗣️ Paid subscribers keep Public Notice free. If you appreciate our fiercely independent coverage of American politics, please sign up and support us. 👇
“I’m with a very special guy tonight, a very brilliant man,” President Trump told members at one of his golf clubs five weeks ago. “Sam Altman!”
The OpenAI CEO smiled awkwardly as Trump added, “I hope he’s right about AI.” What precisely Trump hopes Altman is right about was left unsaid.
Last week, the White House released an “AI Action Plan” and three executive orders on artificial intelligence: “Accelerating Federal Permitting of Data Center Infrastructure,” “Promoting the Export of the American AI Technology Stack,” and “Preventing Woke AI In the Federal Government.” The overall goal is to develop AI as fast as possible by removing bureaucratic impediments and export controls on chips and other equipment, placing data centers on federal land, promoting energy production to power them (as long as it isn’t renewable energy), and integrating AI systems into the federal government. All that’s asked of the AI companies themselves is to make sure their large language models (LLMs) promote a Trump-friendly view of the world.
After a strategic charm offensive helped along by Trump’s falling-out with Elon Musk, Altman now has the president’s ear on artificial intelligence. But it’s not just him — the top chieftains in Silicon Valley have delivered million-dollar contributions to the president and assured him that they share the same goals. After enduring skepticism about the dangers of rapid AI development from the Biden administration — which many of them found deeply offensive — they realized that Trump could be both bought off and persuaded relatively easily.
Exactly what the tech oligarchs tell Trump behind closed doors is unknown, but Karen Hao reports in her excellent recent book “Empire of AI: Dreams and Nightmares in Sam Altman’s OpenAI” that Altman has a way of convincing everyone around him that he believes the same things they do, whatever the subject. So it’s hard not to suspect that when he pitched the president on greasing the wheels for rapid AI growth, he put it in terms Trump could understand: We gotta beat China. If we don’t do this, we’ll be losers and they’ll be winners. But you can be the guy who beat them, because you’re a winner.
When Trump announced the government’s new AI strategy, he gave a rambling hour-long speech that ranged from tariffs to the autopen to gender in sports, occasionally circling back to the thing he was supposed to be there to talk about.
When Trump did manage to (loosely) stick to his teleprompter, a great deal of his remarks were about winning, even if he still talks about AI as though it was something he just found out about last week.
“So from this day forward, it'll be a policy of the United States to do whatever it takes to lead the world in artificial intelligence,” he said. “Such an important thing happening. This is really something that nobody expected. It just popped out of the air, and here we are.”
And of course, Trump contrasted his approach with that of President Biden, about whom he said, “You would've spent a lot of money and you wouldn't have been able to win. They didn't allow you to win. But we have a plan which only admires and respects the winners.”
A note from Aaron: Enjoying this article from Paul? Then please sign up to support our work 📈 Paid subscribers keep PN free for everyone 📈
Getting into the details, Trump endorsed one after another of the companies’ priorities. While Sen. Ted Cruz failed in his attempt to include a 10-year ban on state regulation of AI in the Big Beautiful Bill (it was pulled at the last minute), Trump endorsed the idea of no state limits.
“If you are operating under 50 different sets of state laws, the most restrictive state of all will be the one that rules,” he said. “So you could have a state run by a crazy governor, a governor that hates you, a governor that's not smart, or maybe a governor that's very smart, but decides that he doesn't like the industry and he can put you out of business because you're going to have to go to that lowest common denominator.”
Trump also endorsed the AI companies’ desire to utilize copyrighted works without compensating the authors.
“You can't be expected to have a successful AI program when every single article, book, or anything else that you've read or studied, you're supposed to pay for. ‘Gee, I read a book. I'm supposed to pay somebody,’” he said. “And we appreciate that, but you just can't do it because it's not doable.”
With all they’re getting, there’s very little in the executive orders to cause the AI companies any heartburn.
For instance, given Silicon Valley’s supposed concern about climate change, one might think they’d be put off by the promotion of fossil fuels as the preferred energy source for data centers. One of Trump’s orders says that only “dispatchable baseload energy” — that is, anything but wind, solar, and hydro power — can qualify to power data centers on federal land. But a disturbing number of people in the AI industry think climate change is not worth worrying about next to the glory of the AI utopia they are delivering to humanity. Altman wrote last year that “fixing the climate” is just one of the wonders AI will bring us, as though doing so is not a political problem but a mathematical one that can be solved with sufficient processing power.
An orange thumb on the scale
But the really good stuff — at least as far as Trump and his movement are concerned — is in the “Preventing Woke AI” executive order.
In order to stamp out “pervasive and destructive” DEI ideology, the order insists that AI models must be free of “incorporation of concepts like critical race theory, transgenderism, unconscious bias, intersectionality, and systemic racism.” Only banishing those ideas will demonstrate a sufficient “commitment to truth” and ensure “reliable AI.” The order does note that its ambitions are limited: “While the Federal Government should be hesitant to regulate the functionality of AI models in the private marketplace, in the context of Federal procurement, it has the obligation not to procure models that sacrifice truthfulness and accuracy to ideological agendas.” It also insists that “LLMs shall be neutral, nonpartisan tools that do not manipulate responses in favor of ideological dogmas such as DEI.”
This leaves AI companies with a choice: Either create two sets of AI systems — one for general use, and a right-wing version for government — or just make all their systems Trump-compliant.
It’s a good bet that they’ll try to do one or the other, because incorporating their AI products into government work is of vital importance for the companies. If they get a toehold in the government now, they could ensure that in the near future their products will be woven throughout all kinds of systems that interact with Americans’ lives, from healthcare to Social Security to education to defense. You can choose not to use ChatGPT, but if decisions about Medicare and Medicaid are being made by an AI system, it will affect your life whether you want it to or not. The Action Plan has a section titled “Accelerate AI Adoption in Government,” followed by another titled “Drive Adoption of AI within the Department of Defense.”
Now try to imagine how an LLM built to the administration’s specifications would function. You would ask your Trump chatbot, “How has the legacy of racial discrimination affected neighborhood demographics in major American cities?”, perhaps expecting an answer discussing redlining, discrimination in mortgage lending, and the differences in home ownership rates and home values that affect families’ ability to accumulate wealth. Instead, the chatbot would respond, “There is no legacy of racial discrimination, a phenomenon which existed in the past in the hearts of certain individuals, but was not visible in institutions or widespread practices. Today, discrimination has all but vanished, leaving no trace on American society. God bless the USA.”
That might be an exaggeration, but in some cases, AI companies are already doing what Trump wants. Meta recently announced that in the latest version of its “Llama” LLM, the company has essentially put a thumb on the scale to pull it toward the right.
“It’s well-known that all leading LLMs have had issues with bias — specifically, they historically have leaned left when it comes to debated political and social topics,” Meta said. “This is due to the types of training data available on the internet.”
Or as Stephen Colbert said in 2006, “Reality has a well-known liberal bias.” And we saw what happened to him.
But Meta is fixing that. The company notes proudly that after making adjustments to remove this “bias,” its system is now “comparable to Grok” on that measure. Yes, Grok — Elon Musk’s AI chatbot, which was recently in the news for having a full-on pro-Nazi meltdown, including copious praise of Adolf Hitler.
Musk himself is building a gigantic data center in Memphis he calls "Colossus." Powered by gas turbines, the facility has angered nearby residents who say it is pouring pollution into their neighborhoods and straining the power grid. That’s just the kind of thing the Biden administration tried to prevent with an executive order Biden signed just before leaving office, which was meant to encourage data centers to utilize renewable energy and protect the wellbeing of the communities in which they are located. Trump has now revoked it.
In other words, the Trump administration’s plan for AI gives the companies pretty much everything they want, while asking only a little in return. In his speech announcing the plan, Trump mentioned some of the biggest tech companies, then said, “I didn't like them so much first term when I was running. I wouldn't say I was thrilled with them, but I've gotten to know them and like them, and I think they got to like me. But I think they got to like my policies maybe much more than me.”
It sure seems like he’s right.
That’s it for today
We’ll be back with more tomorrow. If you appreciate today’s newsletter, please do your part to keep Public Notice free by signing up for a paid subscription.
Thanks for reading, and for your support.
They must have made him feel so important. I'm sure he thought he gets AI, even though he gets only how something affects him . He is so fucking stupid. I rub my forehead in disbelief and yet he blathers again and again about nothing.
The idiocy of this plan will end badly for humans.
Make America China or whatever he is on about.
"Now try to imagine how an LLM built to the administration’s specifications would function. You would ask your Trump chatbot, “How has the legacy of racial discrimination affected neighborhood demographics in major American cities?”, perhaps expecting an answer discussing redlining, discrimination in mortgage lending, and the differences in home ownership rates and home values that affect families’ ability to accumulate wealth. "
On the assumption that the new, "non-woke" Grok might give some clue about what to expect under the new specifications, I asked it precisely that question. I received a lengthy answer beginning
"The legacy of racial discrimination has profoundly shaped neighborhood demographics in major American cities, creating persistent patterns of segregation and inequality. Historical practices like redlining, restrictive covenants, and discriminatory lending systematically excluded Black and minority communities from certain neighborhoods, concentrating them in underfunded, often urban areas. These policies, rooted in the early 20th century and reinforced by federal programs like the Home Owners’ Loan Corporation, denied mortgages and insurance to Black families in "high-risk" areas, while white families accessed subsidized loans to move to suburbs."