- More about AI, and how (from Steven Levy) there’s no such thing as a bias-free AI;
- Speculation on what a MAGA ideologically correct AI would be;
- Following up on Harari: how would an AI handle inconsistencies among, and within, religious texts?
- How the administration is hiding information about the USAID cuts;
- Why MAGA evangelicals are obsessed with “sexual deviants”;
- And why do people accuse powerful women of being men?
There’s much about AI that I haven’t explored, but my understanding is that the models, the various Chat apps and so on, are, as Paul Krugman wrote a couple days ago, “a souped-up version of autocorrect.” I’ve seen it in Microsoft Word for two or three years now. As you type into a Word document, the software “predicts” which words you will type in next, on the bases of probable word combinations from the vast amount of data (including newspapers, magazines, and books) in the context of the words you’ve just typed. It works fairly well fairly often. The predicted words show up in gray past the location of your cursor, and if you want to accept those works, you just hit tab. OTOH, it hasn’t helped me from having to type the entire word “totalitarianism” into my notes several times today, as I’m reading the Harari book.
An AI is not a fount of fixed wisdom. Or any kind of truth. It’s feeding you answers based on what it thinks you want to hear.
So who’s to determine what’s biased or not?
Wired, Steven Levy, 25 Jul 2025: Trump’s Anti-Bias AI Order Is Just More Bias, subtitled “The Trump administration says it wants AI models free from ideological bias, as it pressures their developers to reflect the president’s worldview.”
On November 2, 2022, I attended a Google AI event in New York City. One of the themes was responsible AI. As I listened to executives talk about how they aligned their technology with human values, I realized that the malleability of AI models was a double-edged sword. Models could be tweaked to, say, minimize biases, but also to enforce a specific point of view. Governments could demand manipulation to censor unwelcome facts and promote propaganda. I envisioned this as something that an authoritarian regime like China might employ. In the United States, of course, the Constitution would prevent the government from messing with the outputs of AI models created by private companies.
This Wednesday, the Trump administration released its AI manifesto, a far-ranging action plan for one of the most vital issues facing the country—and even humanity. The plan generally focuses on besting China in the race for AI supremacy. But one part of it seems more in sync with China’s playbook. In the name of truth, the US government now wants AI models to adhere to Donald Trump’s definition of that word.
He quotes Trump’s executive order. Then:
That’s all fine until the last sentence, which raises the question—truth according to whom? And what exactly is a “social engineering agenda”? We get a clue about this in the very next paragraph, which instructs the Department of Commerce to look at the Biden-era AI rules and “eliminate references to misinformation, Diversity, Equity, and Inclusion, and climate change.” (Weird uppercase as written in the published plan.) Acknowledging climate change is social engineering? As for truth, in a fact sheet about the plan, the White House says, “LLMs shall be truthful and prioritize historical accuracy, scientific inquiry, and objectivity.” Sounds good, but this comes from an administration that limits American history to “uplifting” interpretations, denies climate change, and regards Donald Trump’s claims about being America’s greatest president as objective truth. Meanwhile, just this week, Trump’s Truth Social account reposted an AI video of Obama in jail.
This is indeed a fundamental problem, as the Harari book is describing. Who gets to claim the truth. Are videos on Truth Social the truth? Obviously not. They want to exclude everything DEI but also climate change as somehow being ideological biases. They are deluded.
But it’s fun to imagine for a moment what a MAGA ideological correct AI would be, or say. (There’s already a conservative analogy to Wikipedia with Conservapedia (this is a Wikipedia link).) What does MAGA believe? The literal inerrancy of the Bible? The falseness of evolution? The supremacy of the white race? Presumably they would like an AI that endorses those positions. As being “non”-ideological.
The problem is, the way AI, or at least those LLMs work, is they consolidate all the data they can find, and build a network of consistencies. It provides the words in Word that most people would type after the words you’ve typed. If you ask ChatGPT a question, it gives you a clean paragraph based on the consensus of discussions about that topic to be found everywhere out there on the web.
\\
Recalling the Harari video I linked yesterday, which concerned how well an AI would know, in a way no individual human can, all sorts of religious texts. The obvious follow-up to these topics is, how do these “AI”s handle inconsistencies? If you feed an AI both the Bible and the Quran and ask certain questions? I’m guessing it would couch its answers with “According to…” this or that. Even with the Bible by itself, which contains hundreds of contradictions and inconsistencies, as folks since Thomas Paine have noted. Would an AI resort to “According to…” this book and chapter in the Bible, and follow up with, “While according to…” this other book and chapter in the Bible… Because waffling about sources and authorities isn’t actually providing any kind of truth. Except a truth about human nature.
\\\
Again, the Harari book I’m about half-way through, NEXUS, is fascinating in an unexpected way. It’s about information, and how different government and cultures can be understood as employing different kinds of information networks. E.g., to simplify greatly, totalitarian governments want all information centralized through a single hub, while the government in turns wants to micromanage how people conduct their business and leads their lives; while democratic governments, with their freedom of information, are comprised of many information hubs. It’s unexpected because the book was published last Fall, and apparently finished just earlier in 2024, in either case before Trump became president again and is implementing so much of the totalitarian playbook. Almost every item in the daily news seems to be part of this. Today’s example:
Vox, Kelsey Piper, 24 Jul 2025: We don’t know how many people will die because of Trump’s USAID cuts, subtitled “The chaos of Trump’s global health cuts make the human toll near-impossible to calculate. That’s by design.”
Centralized control of information, or suppression of it.
\
There are also recurring themes among the conservative populists.
JMG, 25 Jul 2025: MAGA Evangelical Leader Who Warned Of “Sexually Deviant” LGBTQs Pleads Guilty To Child Porn Charges
There are one or two stories like this every week. Evangelical obsession with the sexual perversion of others is a dead give away.
\
PolitiFact, Madison Czopek, 24 Jul 2025: Beyond the Macrons’ lawsuit against Candace Owens: Why do people accuse powerful women of being men?
Hmm, should I speculate, or just read the article?
Owens’ claim is one example of a wider trend of conspiracy theories targeting prominent women in politics and culture, including former first lady Michelle Obama and former Vice President Kamala Harris. Conspiracy theory researchers say that’s because these women are influential, politically left-leaning and break gender stereotypes at a time when the conservative movement favors traditional gender roles.
Well, yes, that’s basically it.
