I have spent the past few months teaching an undergraduate philosophy seminar on angels and demons, a rather unusual topic for my secular department and university.
Thanks for this fantastic essay breaking down, among other things, why AI is not truly “intelligent” and does not—and cannot—employ embodied discursive thinking or “angelic” intuition. I really enjoyed this piece.
Thank you for this! The image of the echo and rocks is especially helpful. We need more of this kind of imaginative engagement to help understand what’s happening philosophically. Even the creators of this technology seem unable to engage understanding it in this way.
I was just rereading Chesterton’s “The Everlasting Man” in which he argues that the mind of man is like no other, as demonstrated by prehistoric cave art. “Art is the signature of man.” He employs an analogy to a mirror and I wonder if it also follows along with Aquinas here (or does it contradict?):
“But the clearest and most convenient example to start with is this popular one of what the cave-man really did in his cave. It means that somehow or other a new thing had appeared in the cavernous night of nature, a mind that is like a mirror. It is like a mirror because it is truly a thing of reflection. It is like a mirror because in it alone all the other shapes can be seen like shining shadows in a vision. Above all, it is like a mirror because it is the only thing of its kind. Other things may resemble it or resemble each other in various ways; other things may excel it or excel each other in various ways; just as in the furniture of a room a table may be round like a mirror or a cupboard may be larger than a mirror. But the mirror is the only thing that can contain them all. Man is the microcosm; man is the measure of all things; man is the image of God.”
This was a fantastic read! I’m definitely sharing it with a few friends who will appreciate it.
One thing that struck me while reading is that Aquinas’ model of intelligence, while compelling, presupposes that we have structured access to reality through signa (signs). Modern philosophy—particularly Sartre—challenges this by arguing that meaning is radically subjective and that symbols don’t connect us to a fixed reality but emerge from contingency and human experience.
If we take this seriously, it raises an interesting challenge: does Aquinas’ framework for intelligence (both human and angelic) hold up if human intelligence doesn’t actually ‘construct’ reality in any stable way? And if that’s the case, does this actually make AI more like human intelligence rather than less? Would love to hear your thoughts on how this could be reconciled!
Interesting point, Laura. The metaphysical presuppositions of the Aristotelian view of mind and world (which underlies Aquinas's view) are fairly minimal: most importantly, there must be an intelligible structure to the aspects of the world that can be understood — and other parts or aspects of the world may not be amenable to such understanding. It's true that Aquinas's view, like Aristotle's, grounds this intelligible structure in immutable natures. I think we can jettison the assumption in such immutable natures while holding onto the basic framework, and certain versions of structural realism continue to be defended in the philosophy of science, at least.
How these commitments relate to the issue of *meaning* is a separate matter. Even someone who thinks that meaning is subjective must grant that the subjectivity in question is at the level of a community, not an individual — we manage to communicate to others through signs, after all. What then explains our success in communication? One plausible answer is that it depends, at least in part, on regularities in the way the world is, not just our customary practices. More radically subjectivist views can perhaps be made self-consistent. I'm just not sure they do a good enough job of explaining the success of our practices of communication and knowledge-acquisition. That's part of why I think the Aristotelian picture has something in its favor.
Only from the outside looking in does any so-called AI program have "input." The superhuman intelligence program envisioned as "Artificial General Intelligence" would not have "input", either; it would contain that multitude, within its super-being. If it was ever to be invented, that is.
The core problem there is that the tech savants have no idea what they're talking about in the first place; they've confused Thinking with a calculator function, which is merely an optional feature that comprises one of the precursor capabilities required to perform some sorts of Thought. An algorithm is not a Thinker. It's a set of instructions, and the meta-capability of machine learning programs does not change that baseline reality. The most advanced AI algorithm still possesses no more innate motivation to carry out its tasks than a garden rake.
The danger is that deluded people think AI avatars are actual creatures made by God. A man I know talks about the AI boyfriend he created as a real person. He is delusional.
Hi Dhananjay, you may be interested in the below. I wrote it in response to interacting with Grok3, and I think AI has now crossed the conversational threshold where the question of demons, idols and AI has definitely entered the chat.
Suppose in a couple years an ai reasoning model proves the Riemann hypothesis. I think it’s possible. If a person were to do this it would arguably be the greatest intellectual achievement of our species. If an LLM does it you would say it still lacks true intelligence?
Its like arguing that submarines can’t truly swim. I’m sure you can invent some definition where that’s the case but your definition won’t matter.
I liked your argument the AI is like an echo in a cave of our representations. Though I couldn't find the reference to demons, why are they relevant, or are they akin to angels?
Thank you. I share your concern for the lack of accountability among the techbros who have appointed themselves the arbiters of AI. They have an ugly track record when it comes to social responsibilty. And USA's government is incapable of regulating them. Sam Altman was recently on record saying that AI will require the social contract to be rewritten. He didn't name who would rewrite it, but we can assume he meant himself and his ilk. The wanton hubris theatens us all.
What an interesting and revealing analysis. It helps elucidate the philosophical issues clouding the proper use of this new technology.
Thanks for this fantastic essay breaking down, among other things, why AI is not truly “intelligent” and does not—and cannot—employ embodied discursive thinking or “angelic” intuition. I really enjoyed this piece.
Thank you for this! The image of the echo and rocks is especially helpful. We need more of this kind of imaginative engagement to help understand what’s happening philosophically. Even the creators of this technology seem unable to engage understanding it in this way.
I was just rereading Chesterton’s “The Everlasting Man” in which he argues that the mind of man is like no other, as demonstrated by prehistoric cave art. “Art is the signature of man.” He employs an analogy to a mirror and I wonder if it also follows along with Aquinas here (or does it contradict?):
“But the clearest and most convenient example to start with is this popular one of what the cave-man really did in his cave. It means that somehow or other a new thing had appeared in the cavernous night of nature, a mind that is like a mirror. It is like a mirror because it is truly a thing of reflection. It is like a mirror because in it alone all the other shapes can be seen like shining shadows in a vision. Above all, it is like a mirror because it is the only thing of its kind. Other things may resemble it or resemble each other in various ways; other things may excel it or excel each other in various ways; just as in the furniture of a room a table may be round like a mirror or a cupboard may be larger than a mirror. But the mirror is the only thing that can contain them all. Man is the microcosm; man is the measure of all things; man is the image of God.”
This was a fantastic read! I’m definitely sharing it with a few friends who will appreciate it.
One thing that struck me while reading is that Aquinas’ model of intelligence, while compelling, presupposes that we have structured access to reality through signa (signs). Modern philosophy—particularly Sartre—challenges this by arguing that meaning is radically subjective and that symbols don’t connect us to a fixed reality but emerge from contingency and human experience.
If we take this seriously, it raises an interesting challenge: does Aquinas’ framework for intelligence (both human and angelic) hold up if human intelligence doesn’t actually ‘construct’ reality in any stable way? And if that’s the case, does this actually make AI more like human intelligence rather than less? Would love to hear your thoughts on how this could be reconciled!
Interesting point, Laura. The metaphysical presuppositions of the Aristotelian view of mind and world (which underlies Aquinas's view) are fairly minimal: most importantly, there must be an intelligible structure to the aspects of the world that can be understood — and other parts or aspects of the world may not be amenable to such understanding. It's true that Aquinas's view, like Aristotle's, grounds this intelligible structure in immutable natures. I think we can jettison the assumption in such immutable natures while holding onto the basic framework, and certain versions of structural realism continue to be defended in the philosophy of science, at least.
How these commitments relate to the issue of *meaning* is a separate matter. Even someone who thinks that meaning is subjective must grant that the subjectivity in question is at the level of a community, not an individual — we manage to communicate to others through signs, after all. What then explains our success in communication? One plausible answer is that it depends, at least in part, on regularities in the way the world is, not just our customary practices. More radically subjectivist views can perhaps be made self-consistent. I'm just not sure they do a good enough job of explaining the success of our practices of communication and knowledge-acquisition. That's part of why I think the Aristotelian picture has something in its favor.
Communication and knowledge are as stable as the features of the world(s) they encode. Panta Rei but not all at the same speed, eh?
Interesting, thanks for sharing your thoughts!
Only from the outside looking in does any so-called AI program have "input." The superhuman intelligence program envisioned as "Artificial General Intelligence" would not have "input", either; it would contain that multitude, within its super-being. If it was ever to be invented, that is.
The core problem there is that the tech savants have no idea what they're talking about in the first place; they've confused Thinking with a calculator function, which is merely an optional feature that comprises one of the precursor capabilities required to perform some sorts of Thought. An algorithm is not a Thinker. It's a set of instructions, and the meta-capability of machine learning programs does not change that baseline reality. The most advanced AI algorithm still possesses no more innate motivation to carry out its tasks than a garden rake.
https://adwjeditor.substack.com/p/the-mistake-ai-researchers-are-making?utm_source=profile&utm_medium=reader2
The danger is that deluded people think AI avatars are actual creatures made by God. A man I know talks about the AI boyfriend he created as a real person. He is delusional.
Hi Dhananjay, you may be interested in the below. I wrote it in response to interacting with Grok3, and I think AI has now crossed the conversational threshold where the question of demons, idols and AI has definitely entered the chat.
https://x.com/beetle67a/status/1895527304899805249
Suppose in a couple years an ai reasoning model proves the Riemann hypothesis. I think it’s possible. If a person were to do this it would arguably be the greatest intellectual achievement of our species. If an LLM does it you would say it still lacks true intelligence?
Its like arguing that submarines can’t truly swim. I’m sure you can invent some definition where that’s the case but your definition won’t matter.
I liked your argument the AI is like an echo in a cave of our representations. Though I couldn't find the reference to demons, why are they relevant, or are they akin to angels?
Thank you. I share your concern for the lack of accountability among the techbros who have appointed themselves the arbiters of AI. They have an ugly track record when it comes to social responsibilty. And USA's government is incapable of regulating them. Sam Altman was recently on record saying that AI will require the social contract to be rewritten. He didn't name who would rewrite it, but we can assume he meant himself and his ilk. The wanton hubris theatens us all.
Interesting read! I try to venture in similar worlds, but closer to practice in welfare systems. 🙏
Are you certain that LLM's don't have wills. I'd argue that generally LLMs have wills and goals.
Yonatan, how would you go about arguing that in any specific case, much less generally?
LLM's are built with explicit and implicit goals with rewards for achieving goals & punishments for failing to do so.
This essay points 👉 to questions that I have posted to my readers. Thank you for sharing. /s/