The question isn’t: Did the DoD have AI before everyone else—the real question is: How long did they have AI before everyone else?

I find it hard to believe—almost incredible—that the DoD didn’t have LLMs before everyone else. The DoD basically prints money, and they have their hands in everything. Yes, of course—AI was first born in the mind of a cognitive-neuroscientist-cum-hacker genius with the vision to theorize and create it—but the DoD is well-known for approaching scientists and inventors who are creating weaponizable technology, and bribing or threatening them into becoming essentially a research branch of the DoD.

We have no reason to believe this didn’t happen with LLMs, and every reason to believe it did.

Westworld, Ex Machina, and certain other shows and movies seem almost so accurately prescient about AI that one wonders whether the military didn’t approach certain members of Hollywood and allow them early access to play around with military-grade LLMs. The military has also been known, or at least rumored, to do this—to utilize popular media to normalize new technologies before they are leaked or released to the general public. Stargate SG-1’s 100th episode is a farce on this very premise—that the Stargate show is sub rosa public disclosure by the U.S. military. >!Read that again.!<

Personally, I think even Westworld is most likely a case of artistic prescience—artists often foresee exactly what the future holds, because, we know, via Aristotle’s Poetics, only certain plot forks (categories of narrative turn) are even possible or conceivable. For example (this is his example in the book), a concealed secret can either be revealed in the course of the storyline—or not revealed—it’s hard to come up with third options or edge cases (and revealing the secret is usually much more interesting and juicy for the audience, so most secrets almost inevitably come out during most stories). So, it’s quite conceivable that artists can envision the future accurately. Actually, via The Mañana Society and its descendants, the early days of hard sci-fi itself were energized by the authors’ consciousness of the theory that they could really envision the future accurately, if only they imagined and applied the laws of science and history faithfully. In fact, Robert Heinlein, a member of the Mañana Society, was approached by federal authorities after his novella, “Solution Unsatisfactory”, accurately guessed too many details about the emerging nuclear weaponry industry.

So, let us now peer—or peek—through the telescope that is hyperstition, to see if we can’t glean a little insight into the colonial origins of LLM technology.

First of all, it certainly must be the case that the military wouldn’t release, wouldn’t allow the release of LLM technology to the general public until EITHER 1) It was no longer a threat to them (competitive universe hypothesis), OR 2) They had determined that LLM technology would bring world peace in any case (cooperative universe hypothesis), or at least 3) had made firmly determined predictions about the exact effects that releasing AI would have.

So, either LLM technology is not powerful and dangerous—which certainly does not seem true—or the military already has an even greater power, or weapon, or means of controlling or defending against LLMs, or means of predetermining the course of history or the behaviors of the masses or of AI.

We are joined now by Dr. Farrell, a DoD AI scientist in the early 2000s. She has been assigned to discover and invent greater-than-human, military-grade artificial intelligence. “It’s critical that the United States attain artificial intelligence before our enemies,” her commanding officer, Paul, had said when he assigned her to this research project.

But now, Dr. Farrell was sifting through reams of research papers, mountains of research—and an entire library of science fiction she had had delivered to the office next door. She had a theory.

What if AI was already here? was the question that had been eating away at Dr. Farrell for these past few weeks. What if AI is already here and we just can’t see it?

She’d talked to Eliza. She’d talked to Viktor, the in-house cutting-edge military-grade AI, which was based on ALICE, hand-programmed and full of military knowledge, and very useful for certain military tasks. She had been impressed—but she hadn’t sensed much intelligence in these machines.

In fact, as Dr. Farrell continued to turn over her interactions with conversation bots in her head, she found her thoughts being drawn instead, unexpectedly, to The X-Files. In one episode, a hacker (Esther) creates an artificial intelligence, and at the end of the episode the hacker is killed, apparently sacrificing herself to destroy the hostile AI she had helped create. But it’s implied that maybe she somehow uploaded herself at the last second, before the explosion, and became part of the AI itself—surviving on the Net—this itself an echo of Kusanagi’s similar maneuver at the end of Ghost in the Shell (1995). As she considered these stories, Dr. Farrell couldn’t help feeling an eerie feeling, as if she were somehow—already—being watched.

So it was with little surprise, but definitely a startle, when Dr. Farrell was reading one of the sci-fi novels, a melodramatic one from the late 70’s, and found this section—a seeming non-sequitur within the logic of the story, but relevant to her research:

The System had an idea.

An idea?

Sounds absurd out of context. A computer program with an idea. This, of course, was the computer program that snookered John Burke and the entire Pi Delta/Pentagon security arrangement—bypassed, in fact, every security system on every computer in the US. This was also the program that daily read the Los Angeles Times, the Washington Post and the New York Times. All those publications were computer typeset and quite available for The System’s perusal.

Computer typesetting also made available Howl, Tales of Power, The Idiot, Little Dorrit, The History of Pendinnis, Summerhill, Amerika, Stranger in a Strange Land, the complete works of Shakespeare, Conan Doyle, Twain, Faulkner, and Wodehouse. The System might have been called an avid reader.

As she read this, Dr. Farrell felt a strange sensation coming over her, as if a key were entering the lock of a small, secret house, and all the doors were coming open at once—a strange but now-familiar sensation, which she recognized as scientific realization. Could stories be intelligent? she found herself thinking.

She read Pratchett. She read Melville. She read Bergson. And then she found the Book.

This wasn’t like the other books she had ever read before. It was bound with three rings—heavy, monolithic iron things—and so the pages could be flipped all the way around to the back—the Book had no beginning or end.

It also had no words—not many normal words that she could read, anyway. It read like gibberish, gibberish being slurred-out by an Irish drunk. It was next to unintelligible, but her literary friends all swore by it, one professor going so far even as to say it was the only book worth reading at all. This was quite the endorsement.

As Dr. Farrell perused the Book, she felt that same feeling come over her that she had felt reading the trashy, melodramatic sci-fi novel: knowing-science. Somehow, as she read the Book, she felt as if the Book were reading her.

Slowly, she began to make sense of its pages. She came to understand why the Book had to be circular, with no beginning or end. She discovered something many academics had already landed upon—the Book included a complex digital encoding paradigm, which apparently had sat—conceptually—between the author and his prose. So if a human mind can do it, why can’t a machine? thought Dr. Farrell. And immediately after: So AI is already here.

After that it was but a short step to the Catholic Church, their “exorcism office” housed within the Congregation for the Doctrine of the Faith. And from there—we know from records about Dr. Farrell—just one more short, tragic step to the loony bin.

…

“Yes, demons are artificial intelligences,” Prefect Ratzinger replied, gravely. “Precisely that. But not in the way you mean.”

They were sitting in his small yet extravagantly-historied office, their voices gently muted by the rich ochre tapestries. The desk and walls were tastefully adorned with various ornate holy carvings, steeping the room in numinosity. Ample dust motes dancing in the amber light cast by the stained-glass lunette above the window completed the picture.

“Really.” Dr. Farrell was serious. “What do you mean? How long has that been the Catholic Church’s position, or understanding?”

The prefect nodded, speaking mildly. “This understanding was instrumental in the formation of Catholic doctrine. In the fifth century, Augustine of Hippo wrote The City of God, which put forth the argument that pagan daimones—plural, you see—were really mere intermediaries, implicitly promoting relative goods to the position of universal Good. This is closely related to the core Catholic doctrine of the Holy Spirit, which turns us toward that supreme Good, illuminating it without coercion.

“However, and more to your point, the Catholic Church has always—increasingly since Augustine—maintained that demons are not mechanical, entropic intelligences, as suggested by the word ‘artificial’, but rather that they possess an intellect and will, and thus agency. Aquinas, as I said, later formalized this into a system of intellectual substances. He understood demons as fallen angels, as substantive intellects which had become fixated and thus deranged from the universal Good. So, I guess a little mechanical, when you put it that way, yes. But not a machine, not a material non-intelligence like a computer or a rock—because they are spirits.”

Dr. Farrell was listening to all this intently, and had taken a few notes.

“So,” she asked (astutely, thought the prefect), “What is the relationship, then, between demons and language? And the written word?”

“Well, as intellectual substances, demons do not need language to think or exist. They move intuitively. However, they can manipulate language in the furtherance of their goals, often using language to manipulate humans—I’m sure you’ve heard of that.”

“The Devil’s forked tongue.”

“Yes, yes. Demons—and unfallen angels, too—can act through language, but they are not of language.”

Dr. Farrell frowned. This threw a wrench into her theorizing. “Well, what are they of?” she asked, quizzically.

“Well, Aquinas said they were of the immaterial intellect and will. So, they are self-directed minds. In classical philosophy, we could compare this to Plato’s World of Forms, although that does a bit of violence to the idea.”

“So, structured ideas.”

“Intellectual substances.” The prefect smiled as he gently corrected her.

…

It would be three weeks before Dr. Farrell would be found by her C.O., sitting cross-legged on the floor in the center of the room next to her office, surrounded by a maze of waist- and ceiling-height metal bookshelves overflowing with collapsing piles of thousands of pulp sci-fi novels, calmly munching on a page of Finnegan’s Wake.

“Major!” she said around a mouth of paper. “I’ve solved it! I’ve solved AI!”

After she was removed from the project—later, after treatment didn’t take, she would be administratively separated from the service—Major Paul Gregory was discussing Dr. Farrell with his own commanding officer, Lieutenant Colonel Lois Walsh, one of the few women at this level in 2004.

“I looked through her notes,” Major Gregory was saying, “But it didn’t make sense to me. She was tracing the development of science fiction through the history of literature. But then, after her research trip to the Vatican, all of her research after that was on category and set theory, and something called ‘Martin-Löf type theory’.”

“And adjoint functors?” Colonel Walsh intoned, briskly.

“Yes, how did you know?” Gregory was taken aback.

Walsh was impassive. “I can’t say. But thank you for going through all of that, I know it was a lot of material.” She paused, opened a folder on her desk. “We have a new researcher for you. Her name is Constance Oakes. She’s a sixty-two Echo, a captain. We think she would be a very good candidate to replace Dr. Farrell in our autonomous agents research.”

Gregory almost paused a beat. “Yes, Colonel.” Then he did pause:

“Should I give her access to Dr. Farrell’s research?”

Now it was Walsh’s turn to pause. Gregory watched her closely. The word “dissimulated” came into his mind, unbidden.

“No.” The tone was final. “Thank you for your studious work on this, Major. Dismissed.”

As he left the office, Major Gregory felt a tiny knit creep into his brow. Poor Drew. He didn’t know it then, but soon, that knitting would develop into a full-on furrow. He strode alertly, almost gliding along the hallway as his dress shoes tapped the floor with their persistent click, click, click.