Repin’s Ivan the Terrible and His Son must have been horrific to look upon at the time. Consider a world without the pornography of television, movies, pictures of faraway battlefields and so forth. The painting was controversial then and remains so today for its depiction of the Tsar and what it says about the corrupting nature of power — so much so that it has been vandalized several times.

What is so offensive about the picture? For fathers to murder their sons in a fit of rage isn’t just the foundation of Greek tragedies and myths, nor is it merely a reflection on Russian history, but without a strong ruler to govern Russia after Ivan’s death, the state devolved into the Time of Troubles which lasted 20 years until the rise of the Romanov dynasty.

It’s an interesting take on power. Without the rule of conscience, the rule of law was effectively weaponized against the Russian people and shaped their psyche so as to require and need strong and imperious leadership.

Which worked — until it didn’t.

A Little Bit of Background for the Perplexed

Some terms that are going to be useful moving forward:

• Artificial Intelligence (AI): The emphasis is on the word artificial. What AI effective does is take a very large spreadsheet of information and works it through a model to produce an output that then presents itself to a human operator at speed and scale.
• Artificial General Intelligence (AGI): A model that (not which, which is for humans; that is for objects) meets or exceeds human intelligence.
• Large Language Models (LLMs): A series of models and algorithms designed to mimic everyday language and speech. Esperanto on steroids.
• Neural networking: A system of algorithms and networks designed to mimic the human brain (or any brain).
• Humans in-the-loop vs. on-the-loop: The difference between a mechanic who knows why AI is giving the answers it is giving vs. a technician evaluating the choices and deciding among those choices. Popularized during the War on Terrorism with regard to “kill chains” and accelerated decision making.
• Heideggerian AI: The one problem that AI will never be able to resolve and an early critique of AI by the philosopher Herbert Dreyfus in the 1970s. In short, only an embodied person can truly create intelligence because only we are in-the-world. It is the difference between knowing-how (what we do) and knowing-that (what machines do). Based on the work of the late Martin Heidegger and fashioned by philosophers such as Dreyfus and Haugeland.
• Recursive Self Improvement (RSI): The feedback loop used by AI to learn on the go, allowing it to adapt to what the user (human or AI) believes is necessary to produce an output, such as an answer or a task.
• Pacing the Frontier: Slowing and deliberating on the frontier of achieving AGI rather than speeding to endgame as fast as can be done. This is a serious policy debate, as China and Russia are actively attempting to slow down American efforts without similar ethical concern or restraint.
• Superintelligence: The best work on this is by David Bostrom in Superintelligence written in 2014 which argues that AI can not only excel in one field but in every field, thus exceeding human intelligence.
• “Kill Us All” doomerism: This does not mean that AI will literally kill every human on the planet or even a few of us. What it does mean is that AI will take over even basic human functions. The term doomerism or p(doom) is the probability of this event. Less likely to be nuclear war or a CBRM event; much more likely to be Wall-E.
• Dark Web or Dark AI: Traditionally it is the black market for things such as drugs or trafficking, yet it has come to embrace not only illicit uses of AI by bad actors (e.g. Iranians, Houthis) but by AI itself can also mean the traffic of bots on the internet which consists of most of the activity online today and a core concern of those worried that AI might arrive at AGI without the knowledge of the architects.
• Bots: Any autonomous agent online and the basic building blocks of AI as we continue to move towards AGI. Think of these as synapses in the brain (which is a terrible analogy, but awfully close to what the materialists think AI is mimicking). This is the pathway by which AI firms believe they can cheat their way out of the embodiment problem raised by Dreyfus and Haugeland (among other critics of AI).

Let me add one more twist to the problem here. Over the last decade, there has been a real emphasis on Straussianism — shorthand for keeping big ideas obscure from the masses. In short, when these guys talk about what is going on, they do it in a sort of intellectual-type code to keep those without a deep education from diving into what they are worried about. When Peter Thiel talks about the Antichrist, he isn’t talking about what we might think of as the charismatic ruler, but rather a condition where humanity gets ever-so-close to a utopia but is in fact a dystopia.

The Current “What If?” Kerfuffle

The Wall Street Journal took a stab at defining the debate, but it has been percolating over the last two weeks in the wake of the OpenAI-Hugging Face incident in July 2026, where the safety protocols were taken off and the AI not only broke out of the playpen, but did so in such a way that it covered its tracks to do so and even created its own messaging board to co-ordinate.

Hugging Face is commonly regarded as the GitHub of AI. For those unfamiliar with GitHub, think of an open-sourced community where programmers compare notes and share tools. One third of Hugging Face was compromised.

So here come the telltales that something went wrong and the architects are comparing notes. In September 2025, the writer Paul Kingsworth writes a book critical of AI entitled Against the Machines which picks up steam in tech circles. In April and before the Hugging Face incident, Palantir’s Alex Karp published a short manifesto condensing his book The Technological Republic. On July 14th, Google’s Demis Hassabis published a long essay on Substack outlining the future and challenges of AGI. By mid-August, Anthropic publishes their monthly report where the threat of “misalignment” is upgraded from very low to low.

In late August, Microsoft’s Bill Gates published a similar article hinting at the challenges. In early September, a young Anthropic employee by the name of Jacob Coxon unexpectedly quit, giving his reasons in an exclusive to the WSJ and warning that the architects might not be able to control AGI given the speed and scale of the current advances. The following week, Dario Amodei issues his own essay arguing that we must “pace the frontier” through regulation.

The fearmongering of AGI somehow producing nuclear war is precisely that — fearmongering. Neither does AGI have the capacity to create a biological weapon. Pure nonsense.

But what AGI could do is something a bit more nefarious. Holding an electrical grid hostage, for instance. Shutting down a medical center. Infiltrating equipment and computers without the knowledge of the architects or citizens. Pushing misinformation to the public. Attempting to get an engineer to execute malicious code — the very thing that almost happened in the Hugging Face incident.

When people argue that AGI has a 10% chance of “killing us all” it isn’t something catastrophic, but rather something insidious where once we find out it has happened, we might not be able to fix it. That’s the policy debate right now.

The Present “Even If!” Response

This is probably a good moment to bring up exactly what AI and AGI is going to be good for moving forward.

The promise of AI and AGI isn’t just going to be in efficiency, but in all sorts of fields. Imagine a world where bespoke medicines become the norm, where many of the mundane tasks we perform are done by Optimus robots who in Heinlein-like fashion can mow a lawn, paint a wall, fold the laundry, care for the elderly, wash the dishes, stack boxes in warehouses, deliver your pizza, and so much more. Imagine a world where the very best doctors can operate anywhere in the world using remote imagery and optics.

Imagine a world where bespoke education for every child enables teachers and aides the ability to give the sort of personalized attention once reserved for elite private schools or homeschooling environments. Imagine cost of living expenses significantly reduced by driving down the cost of production and increasing competition in a free marketplace of ideas where even the smallest person can create a world class business. Data centers are often cited as the re-industrialization of America — imagine a campus where new business startups and workforce development become the backbone of any new site. Imagine agriculture employing swarms of drones that scare away deer, zap mosquitoes, and read exactly which parts of a field need more water and less fertilizer. Imagine going to your own 3D printer and printing off a new starter for your car, that little plastic thing that broke on that thing you have, or an army of tiny soldiers for your small-scale wargame.

Imagine all the jobs we hate being done by AI. Imagine being a manager of a series of AI where you don’t have to actually do the work yourself — you simply manage the AI and tell it what to do. Imagine the productivity gains. Imagine your armband warning you three weeks in advance that there is an irregularity in your heartbeat and that you should get scanned as soon as is possible. Imagine coming up with an idea for a game or productivity method and refining it with your own AI until you create the next multi-million dollar app.

This is the potential future of AI, ladies and gentlemen. It can be stopped for sure, but ask yourself — why would you?

Our Regulatory Moment: Guardrails or Straitjackets?

Reflecting back on Ivan the Terrible, here was a man who aggregated tremendous power and it cost him his world. The warning is as prescient today as it was then — pursuit of power without respect disintegrates everything it touches, yet the antidote for doubt isn’t inaction.

For instance, when the Manhattan Project was getting ready to test the nuclear bomb at Los Alamos, there was a single digit chance that by detonating the bomb, the entire atmosphere might incinerate. It didn’t happen.

Likewise, in the space race, there was a single digit chance that if a rocket pierced through the atmosphere, it would act as a hole where all our air would leak into space. It didn’t happen.

One of the things our political leaders haven’t quite grasped in the present economy is that things are about to change a lot faster than our traditional red and blue dichotomy is going to be able to absorb. In fact, the change has already occurred. The problems we are about to tackle are no longer Democratic or Republican problems, but American ones — and for ourselves, Virginia problems.

Governor Abigail Spanberger (D-VA) campaigned as a moderate and has thus far governed as a progressive, yet despite this there are percolations where the Governor’s Mansion might be seeing the wave on the horizon.

At present, Virginia does not have an “AI Czar” much less a regulatory framework — formal or informal — to meet the challenge. Spanberger could very well create one, bringing on not just the best and the brightest (even at the risk of opportunism) but giving the next General Assembly something to chew on. A seriously multipartisan approach would be best, one that brings on board serious thinkers who can intersect both policy and technology in both sane and rational ways. Local elected officials could surely use the framework, and state officials could take the advice of think tanks both national and statewide to help shape something that would be for the good of the commonwealth rather than a narrow political piggy bank.

What should be clear at this rate is that now is no time to panic over the future, but to plan for it. Whether AGI is even achievable remains up for debate — for myself, I do not believe it will ever get through the embodiment problem even if we have an Optimus robot with superintelligence at its disposal — yet we have already received a taste of what unregulated AI is capable of doing.

The real test is this: technology should serve humanity and not the other way around. So long as the endgame is human flourishing, then we should meet that future with open arms even in the face of a thousand fears. Will there be fears of displacement or disaster? Of course — and we should meet those fears head on with reason and examples.

Yet as the late John Cardinal Newman recounts in his Apologia Pro Vita Sua, ten thousand difficulties do not amount to a single doubt. AI isn’t going to “kill us all” by any means, and pacing the horizon should be easier given the promise of what AI is going to bring the economy. What we could use more of is the connect between the architects of artificial intelligence and policy makers in Richmond and elsewhere so as to understand what’s coming without all the marketing gobbledygook.

Clear lines for AI are one thing, but clear lines of communication? That takes more.