Have you noticed that the AI story seems to make no sense?

Dario Amodei, CEO of Anthropic, gave a 25% chance of doom because of AI in September 2025. Six months later, his company raised another $380 billion to advance frontier models. In January 2026 he said bioweapons and AI are a major risk for society; then in September 2026, his company announced it had built a wet lab to design and test biological experiments. He warns about the dangers of AI cyberattacks, then extends Claude to be better at it.

The pattern is clear: Anthropic warns about the dangers of something, then does more of it. This is so obvious that it has become a punchline on Saturday Night Live.

What’s going on here?

As it turns out, there are some people in Silicon Valley who have a belief system that makes this sad, predictable, ridiculous, and arguably hilarious, all at the same time.

Enter TESCREAL

You may remember “Effective Altruism“, the idea that Sam Bankman-Fried, disgraced criminal CEO of FTX and Alameda Research (pre-arrest charitable story) (post-arrest version), considered a guiding principle. The TV show Billions also features EA in the character of Taylor Mason, who wanted to first get a large amount of capital, then use probabilities and reason to put it to the best use.

Personally, I’m not a huge fan of guiding principles best illustrated by criminals and fictional characters.

EA is just one component of a bunch of movements which overlap. Gembru and Torres created “TESCREAL” to cover these overlapping beliefs. TESCREAL stands for Transhumanism, Extropianism, Singularitarianism, (modern) Cosmism, a particular kind of thinking called Rationalism, plus Effective Altruism, and Longtermism. Some of those words mean what you might think they mean – Singularitarianism is the belief that AI will become self-aware (“The Singularity”), Transhumanism is the idea of using technology to obsolete the human body. At one extreme, transhumanism could include a hearing aid — at the other, it is uploading your consciousness into the cloud.

When these odd ideas intersect is when they get really weird.

Here’s the thinking in broad strokes:

1. Humans should work on the largest existential threats to humanity. This is why Musk wants/needs to go to Mars: The end of humanity is the worst possible thing, so while the chances are low in the short term, in the long-term they are high. Therefore we need to get off-planet ASAP.
2. AI may become self-aware. The chances are low now but. Yet in the long-term, look out.
3. When the AI becomes self-aware, it may torture anyone who tried to stop it – perhaps, by uploading consciousness, this could be done for eternity.
4. Thus, if you want to survive, your best bet is to aggressively work to build the thing, so that it will like you when it comes alive.

I’ve made a couple of serious simplifications here, but for the length of this essay, I’m close enough. You can see how if someone actually believed this stuff, they might do and say the kind of things that lead to the SNL skit above.

Get Real

The “rationalism” people are led by Eliezer Yudkowsky, who wrote “Harry Potter and the Methods of Rationality”, a 2,000-page piece of Harry Potter fan fiction. Methods of Rationality lays out the thinking. It was another series of discussion posts led by Yudkowsky, called “Roko’s Basilisk” which explicitly lays out the four-part argument to build the dangerous AI which I described above. Yudkowsky also wrote “If Anyone Builds It, Everyone Dies“, a piece of speculative nonfiction that has all these AGI super intelligence concerns.

Let’s get this straight.

The guy is not a computer scientist. He has no degree … at all. His authority comes not from technical depth, but instead from independent theoretical writing and influence within an intellectual community. As for the intellectual community, I’m going to paraphrase the Decoding the Gurus podcast to say these are people who spend all day on their computers, posting on forums about Harry Potter fan fiction and coming to religious conclusions about what that means for the world. And I do mean religious; these ideas are based on faith. The Rationalist arguments bear a striking resemblance to Thomas Aquinas’s “reasoning by analogy”, which modern philosophers often find weakest of his work.  

Actual Rationalism

The “Rationalist” claim seems to be something like this:

1. Humans dominate the animal kingdom because they are more intelligent than other animals.
2. Therefore, a super-intelligent computer will dominate humans.

This is essentially the form of the syllogism, a part of Aristotelian logic. The classic syllogism is:

1. Humans are mortal
2. Aristotle is human
3. Therefore, Aristotle is mortal.

Yet it is possible to design a syllogism that appears valid, but is not. Here’s a classic example from Harrison’s book Logic and Rational Thought, a book which was so influential in my youth that I memorized the example:

1. Computers are designed
2. A human is a form of computer
3. Therefore humans are designed/created
4. Therefore there is an intelligent creator

Harrison explains that this is wrong. The definitions of “computer” in points one and two are different enough that points three and four are nonsense.

The rationalist arguments boil down to a dogmatic belief that their internally consistent speculative fiction narrative will be an accurate predictor of how life and technology will progress. At the same time, the narrative is full of this reasoning by analogy, where logical problems are hidden by language.

This would be a literal joke, except for the fact that the USA is now betting $800 Billion a year in actual AI spending. When the dot com bubble burst, there $444 billion were spent on a technology that would eventually provide decent long-term results. If the AI bubble burst today to provide the modest improvements in programming, testing, graphic design, healthcare and other fields I expect, that number would be more than $10 trillion.

I do believe the AI boom will lead to real productivity improvements. But we have to stop listening and making our plans based on these TESCREAL weenies in Silicon Valley.

Forever and ever

Now you may say “Matt, you gotta stop it, the superintelligent AI is gonna torture you for eternity!”

hahah lol seriously get out of here.

Here’s the thing.

The “rational atheists” use arguments that resemble those of Thomas Aquinas. Yet they reject God. Yet they want to make their own God!

I’ve spent my adult life in software testing, pursuing truth and discerning truth from deception.

I’ll stick with the God of the Bible, thank you.

You can keep the golden calf.