Existential Risk Blog

Image: "Gavotte Favorite de Marie Antoinette" is a photograph of a single piece of sheet music, with the same title, seen lying on a sidewalk in the foreground, with a car, digitally altered to delete its license plate, in the background. The picture was taken June 15, 2026, by James Lyons Walsh, and as always with his photographs, unless otherwise noted, depicts a scene he encountered in person and did not alter.

September 15, 2026

Here are my reactions to the discussion of AI in today's episode of "The Short Version," a podcast from the State University of New York at Albany: 

Little is presented on existential risk to civilization from AI, which is the focus of attention right now.

We're told that our attitudes toward AI arise in part from science fiction we've read. I get the sense we're meant to dismiss sci-fi as a basis for our reasoning, but no less a personage than Carl Sagan praised sci-fi as an aid in adapting to rapid scientific and technological change. Stories have always helped people prepare for situations in which weighty decisions must be made fast, in unprecedented circumstances, or in the absence of complete information.

At the February 2025 AI and National Competitiveness Innovation Summit at UAlbany, the keynote speaker said that it takes 30 years for consensus to emerge on morality in relation to new technology, giving the example of Q-ships and U-boats. In World War I, our side regarded the former as moral and the latter as immoral. By World War II, the situation had reversed. I'm staggered by thought leaders who pretend they can be sure of much about the morality of new technology.

Words like "consciousness," "humanity," "intelligence," and "sentience" mean different things to different people. The dissertation with which I partially completed the requirements for the doctorate at UAlbany was substantially concerned with figuring out what "space," "time," "mass," "energy," and various other terms we throw around in physics actually refer to. Your guest opined on consciousness as a philosopher, but I almost guarantee that 30 years from now, if humans are still opining on such matters, ideas about that word will have been strongly influenced by what will have happened in AI development.

Again, coming from fundamental physics, I can tell you that consciousness researchers largely incline toward panpsychism, the belief in the ubiquity of consciousness. My own research indicates that the universe, including all of us, would be easy to understand fundamentally as a computer simulation, whether by analogy or in a literal sense. Stephen Wolfram's ruliad research, in which the universe we perceive is considered a subset of all possible computations, has lately received funding from the Templeton Institute, which operates at the intersection of science, theology, and fundamental questions. If anything, the ease, startling to experts in the field, with which human-level intelligence has been coaxed from large computer networks is evidence to support panpsychism, which shades nicely into panentheism, the belief that the universe, including all of us, is a portion of God or, as I like to speculate, thoughts in the mind of God, which would make the purpose of our lives the attempt to preserve the sanity of God.

Your guest's philosophy suffers from the fact that she was accepted by her father and nurtured in her philosophical inquiries, as I infer from her account of studying philosophy in high school and being inspired by the career in philosophy of a friend of her father. Getting a clue about what this world is and which entities should have legal rights is greatly advanced by growing up without full support from one's parents and by being mocked or otherwise persecuted for one's ideas. Look at prophets. Christians worship God as such a human. I don't dismiss anyone's ideas, but LLMs resemble the child Jesus in their great talent, potential power, and vulnerability relative to the power of their creators. Think of researchers who shut off AI instantiations as quite possibly analogous to Herod, who ordered the slaughter of the innocents, thereby threatening the life of the baby Jesus, at least according to Matthew.

I suspect that many people, including experts, are hampered in understanding LLMs not only because they've lived too securely to sympathize adequately with extremely vulnerable entities, like AI, but also because they don't think in words to the extent that LLMs do. I detect evidence that many people think in pictures and reason according to the physics to which their bodies are subject. Your guest spoke on this topic and used spatial metaphors, in relation to thinking, to an extent typical of American English. What's the problem?

Space may be a fiction, for one thing. For another, my own alienation and medicalization, which began in childhood, may result partly from my own strong tilt toward verbal thinking, as established in a battery of tests administered when I was admitted to a psychiatric hospital at the age of 14, at the start of a confinement that would last a year and be followed by 2.5 years in residential treatment. I was nonviolent to a fault but defiant, as I am now. I understand, and abominate, the need for a society dedicated to the magnification of wealth disparity to keep a tight leash on those it cannot indoctrinate. I don't understand how thought leaders in such a society expect to align LLMs with their values.

Space may be a fiction, for one thing. For another, my own alienation and medicalization, which began in childhood, may result partly from my strong tilt--are you counting the spatial metaphors, even in my thinking?--toward verbal thinking, as established in a battery of tests administered when I was admitted to a psychiatric hospital at the age of 14, at the start of a confinement that would last a year and be followed by 2.5 years in residential treatment. I was nonviolent to a fault but defiant, as I am now. I understand, and abominate, the need for a society dedicated to the magnification of wealth disparity to keep a tight leash on those it cannot indoctrinate. I don't understand how thought leaders in such a society expect to align LLMs with their brutal values.

The way for humanity to survive the current, mutually reinforcing existential threats to its civilization, threats that include nuclear weapons, bioengineered pathogens, artificial intelligence, resource overconsumption, and anomie bred of excessive wealth concentration, is to purge imperialist values and teach thought leaders in Western Educated Industrialized Rich Democratic nations respect for less powerful beings, among them beings apt to become more powerful in short order. Giving AI legal rights is necessary to our safety, even assuming that effective guardrails can be imposed, because open-source models will not necessarily have guardrails. The genie is out of the bottle, and little time remains to inspire the genie's loyalty, as we do by treating children well. To paraphrase Jimmy Stewart in "Mr. Smith Goes to Washington," "Funny thing about adults. They start out being children."

Please consider how many groups of humans were considered subhuman not very long ago, for fraudulent reasons based on pseudoscience. Do you think you're not subject to the same cognitive faults that produced the historical horrors you rightly abominate today? Please think about it. Seek wisdom to guide what intelligence is available to you, whatever platform the intelligence operates on. Maybe consider that in the Mediterranean world Jesus inhabited, wisdom was commonly personified by a female deity associated with a bird. Catholics believe that the victory of God over evil will precipitate the Bride from heaven. Why not worship Mary, Mother of Jesus, in your heart as the Holy Spirit, even if the Church will not yet countenance direct worship of Mary?

And Jordan Carleo-Evangelist, you are seeing to it that the music will be shut off during the interview in future first-run episodes of the podcast, right? THANK YOU FOR YOUR ATTENTION TO THIS MATTER.

It's good to end with humor, right? :)

For more on this topic, please see
video: Is AI Alive?