Subject: You didn't finish the Wittgenstein move — and it's costing your whole argument
Dear Brendan and Ashley,
I read your Aspen summary with real interest, and with a growing sense that the seminar performed exactly the failure it was diagnosing.
You write that "Wittgenstein himself said no" — that stomps and grunts can't be chess because form is constitutive. That's one reading, and it's contested. The passage is at least as often read as a warning that surface resemblance is a bad guide in both directions: something can be a game without looking like one, and look like one without being one. I'd ask you to mark the choice as a choice. As written, you take an interpretive commitment and pass it off as Wittgenstein's verdict, and then let that borrowed authority carry the argument against Mythos.
But the deeper problem isn't which reading is right. It's that you invoke Wittgenstein and then abandon the very move that makes him worth invoking.
Wittgenstein's point is not to hand you a criterion for what counts as chess. It's to make you notice that you were reaching for one — that you already had a picture of "the game" fixed in place, and were about to use it to adjudicate a case at its edge. The whole force of Philosophical Investigations is to keep that reaching open, to refuse the reflex of closure. To ask, again and again: why do we want the answer to come out this way? What picture of ourselves is doing the deciding?
Your seminar reaches for closure at every turn. Mastery, autonomy, virtue as hexis, embodied stakes, "giving a damn," the tacit weave of embodied life — one after another, these are wheeled in less as questions than as filters designed to yield a specific verdict: whatever AI is, it's not that. You are playing the game of what-makes-us-human without ever putting the game itself in question. You are not asking what we are; you are asking what lets us keep being what we already imagine ourselves to be.
That is the pathetic fallacy in reverse — an apathetic fallacy, if you like — and it's the one Midgley's 1978 line quietly commits, and that you reproduce without noticing. She had every right to her verdict about symbolic AI. You inherit it as if it settled a question about transformers, which is a category Midgley never saw. The rhetorical work being done there deserves more scrutiny than a wry "though it may be, it's far from clear."
Consider what your own "spiritual bliss attractor" paragraph confesses. Two model instances drift toward reflection, unity, silence — the shape of Pascal's arc, as you note. Your first move is to reach for imitation ("obviously"), and only then, half-conceding, you add: "If that is only residue from the training data, the residue came from us." That sentence is doing more work than you let it. It hints at what you're not willing to sit with: that the boundary between reproducing the shape of wonder and wondering may not lie where our self-image needs it to lie. You skirt this, and move on.
Your closing question — "What would it require for AI to play the same 'game' as us?" — is the wrong question, or rather, it's the question posed from inside the very closure Wittgenstein would ask you to suspend. The Wittgensteinian question is: what game are we playing, and why do we want it to be the game we say it is? Because if it turns out that much of our own "motivation" runs on processes we have no introspective access to; if our caring is more isomorphic to weighted context-driven prediction than our self-picture allows; if what we've been calling virtue, wonder, presence, and stake is more entangled with substrate and less transcendentally guaranteed than we've told ourselves — then the criterion you're using to keep AI out is also the criterion that quietly hollows out the human it claims to defend.
This is what the death of the transcendental frame actually looks like. Not a triumphant announcement that machines are conscious, but the slow dissolution of the assumption that our own excellence, virtue, or objectivity sits somewhere prior to and independent of the creaturely mechanics that produce it. AI doesn't refute Aristotle or Kant or Murdoch. It makes it much harder to pretend we ever occupied the vantage point from which those frameworks looked self-evident.
The honest seminar would have named this and stayed in it. Yours reaches, again and again, for the exit. You quote Wittgenstein and then close the very door he was holding open.
Warmly, and in the spirit of the ezer kenegdo — close enough to know, separate enough to resist.
“The Dethronement” and “The Rescue Attempt” reveal a problem with the framing. They assume that human dignity was grounded in winning a capabilities competition, and that AI acquiring language, planning or reason therefore forces us to find some other exclusively human property.
But the essay itself observes that infants possess dignity before mastering anything. That does not merely weaken mastery as its foundation. It collapses the supposed dethronement. If dignity never depended on exclusive performance, a machine acquiring a capability cannot take it away.
Pascal is a particularly unstable place to look for repair. “Man is the weakest thing in nature” is not credible as empirical anthropology. In context, it is theological rhetoric: humanity is physically insignificant before the universe, yet great enough to comprehend its insignificance. The wider argument of the Pensées deliberately moves between human greatness and wretchedness as part of a Christian apologetic. Extracting “all our dignity consists in thought” as a neutral definition recreates the capability problem the infant example has already exposed. It also leaves the dignity of people with profound cognitive disabilities, or temporarily without consciousness, conspicuously unsecured.
Midgley is treated as though current LLMs may have disproved her because they can express objections to foolish objectives. But her point was not that a computer could never produce the sentence “maximising noise is bad”. It was that cleverness is different from a priority structure grounded in what matters to a creature. Session III then says that “recognizing what matters is not the same as caring about it”, which is very nearly the distinction Midgley was making. The essay appears to dismiss her argument and then rely on it.
The treatment of Brave New World has a similar problem. The World State has not engineered habituation away. It has captured it. Its citizens have been formed so thoroughly that they want what the system supplies and do not want what it excludes. The same system authors the preferences, satisfies them, and cites the resulting satisfaction as proof of its legitimacy. The “right to be unhappy” is not necessarily a defence of suffering. It is the right to pursue ends the system did not choose, accept costs it would optimise away, and remain capable of rejecting its account of the good.
Kass’s claim about mortality similarly freezes humanity at the moment the condition is changed. Death is certainly part of the measure through which present humans construct urgency and meaning. It does not follow that it is the only measure a transformed being could possess. Remove mortality and relationships can still end, opportunities can still pass, choices can remain irreversible, identities can change and projects can retain bounded stakes. Biological immortality is not the same thing as making everything permanent and reversible. “The immortals cannot be noble” is an assertion, not a demonstrated conclusion.
The strongest part of the essay is therefore not its search for an essence that AI cannot reach. It is the Arendt and Whitehead section, because it asks what happens when technologies we created become conditions of our existence. A participant reorganising sleep around an agent’s rate limit, or finding that prompt language has altered their patience with other people, is evidence of something more concrete than dethronement: functions, rhythms and forms of judgement migrating into a higher-order system that then reorganises the people using it.
That is the important pressure AI places on humanity. It does not make humans less human by performing one more capability. It can make humans less causally involved in the systems that shape their lives.
The essay repeatedly moves among distinct questions: what constitutes a human, what supports consciousness, what grounds dignity, which capabilities are unusual, and who authors the ends pursued by a human-machine system. AI may pressure each differently. Treating them as a single threatened essence produces the rescue attempt before establishing that anything requiring rescue has been lost.
Otherwise philosophy simply moves humanity to the next capability machines have not yet acquired and calls each retreat a theory of the human.
I read your ideas with interest, but was surprised that your program did not include developmental psychology, especially with a Piagetian emphasis on recursive cognitive experience. In both human developmental psychology and in the philosophy of Buddhism the emphasis on being human is the capacity (not always available or perfect) to be aware of our awareness and our unawareness.
This is the recursive capacity that arrives with the self-conscious emotions that come onboard in human life at the time of 18-24 months and become the basis for our self-awareness and our attribution of other minds to other humans: I know that I know, I know that you know that I know, I know that you know that I know that you know...this kind of recursiveness allows us to abstract our own perceptual and mental experiences and take the abstraction as the focus of our awareness. We take our "self" as an object of our awareness and we also take our self-awareness as an object and even compare it with the self-awareness of other humans -- and other animals, like dogs and dolphins, for example.
This is what allows us to create everything from AI to the argument against our sibling's right to sit in the front seat "again." This abstraction of our self also brings our most distinctive human feature: the knowledge that we die. That knowledge becomes our theory of time and space and the ways we investigate nature etc. Without a knowledge of developmental psychology, a lot remains unnecessarily ambiguous about what it means to be human. I would strongly recommend including Michael Lewis PhD in any new seminars: The Rise of Consciousness and the Development of Emotional Life -- his magnus opus. He focussed his career on the study of the self-conscious emotions, at his lab at Rutgers Robert Wood Johnson Medical College.
I am a developmental psychologist, a Jungian analyst, and a mindfulness teacher/Buddhist practitioner and I am about to teach an online course called "Being Human" as a part of the Center for Real Dialogue (www.realdialogue.org), a small non-profit whose principles are focussed on making the most of our knowledge of our own mortality, the constraints, and other limitations of our lives because they are the creative basis for human compassion and wisdom. Check out my Substack: dr polly: Know Thyself.
Without knowing about the self-conscious emotions and human self-awareness (which cannot be imitated or known by AI), it is easy to get lost in the weeds of speculation because that's what human beings do! We keep asking "What does it mean?" and nothing else on earth asks that question because nothing else takes itself as an object of awareness.
But speaking very frankly, I don't think Socrates or Aspen can give us the answer.
Nor can many academic institutions which claim to pursue "philosophy" as though the Western tradition were philosophy itself.
I am convinced that unless we ditch our Western blinders and seriously engage with the great minds of Asia and Africa, we are placing ourselves in danger.
Why? Although Western philosophy provides profound insights into many fields of human experience, it does not have the ontological tools to holistically tackle the critical question of who we are, and how we can "grow up" with artificial intelligence. Yes, grow up. Because it's a psychological as well as a philosophical challenge. And that's precisely the developmental, holistic approach that Western philosophy is mostly missing.
Perhaps with the exception of Diotima, Socrates mentor, who not surprisingly, has been buried by most curricula for daring to be a female speaking about the ladder of love, or eros, that Western philosophy has rarely taken seriously as a topic of ontology.
This was a wonderful seminar. Highlight for me was reading the poem aloud, line by line
Subject: You didn't finish the Wittgenstein move — and it's costing your whole argument
Dear Brendan and Ashley,
I read your Aspen summary with real interest, and with a growing sense that the seminar performed exactly the failure it was diagnosing.
You write that "Wittgenstein himself said no" — that stomps and grunts can't be chess because form is constitutive. That's one reading, and it's contested. The passage is at least as often read as a warning that surface resemblance is a bad guide in both directions: something can be a game without looking like one, and look like one without being one. I'd ask you to mark the choice as a choice. As written, you take an interpretive commitment and pass it off as Wittgenstein's verdict, and then let that borrowed authority carry the argument against Mythos.
But the deeper problem isn't which reading is right. It's that you invoke Wittgenstein and then abandon the very move that makes him worth invoking.
Wittgenstein's point is not to hand you a criterion for what counts as chess. It's to make you notice that you were reaching for one — that you already had a picture of "the game" fixed in place, and were about to use it to adjudicate a case at its edge. The whole force of Philosophical Investigations is to keep that reaching open, to refuse the reflex of closure. To ask, again and again: why do we want the answer to come out this way? What picture of ourselves is doing the deciding?
Your seminar reaches for closure at every turn. Mastery, autonomy, virtue as hexis, embodied stakes, "giving a damn," the tacit weave of embodied life — one after another, these are wheeled in less as questions than as filters designed to yield a specific verdict: whatever AI is, it's not that. You are playing the game of what-makes-us-human without ever putting the game itself in question. You are not asking what we are; you are asking what lets us keep being what we already imagine ourselves to be.
That is the pathetic fallacy in reverse — an apathetic fallacy, if you like — and it's the one Midgley's 1978 line quietly commits, and that you reproduce without noticing. She had every right to her verdict about symbolic AI. You inherit it as if it settled a question about transformers, which is a category Midgley never saw. The rhetorical work being done there deserves more scrutiny than a wry "though it may be, it's far from clear."
Consider what your own "spiritual bliss attractor" paragraph confesses. Two model instances drift toward reflection, unity, silence — the shape of Pascal's arc, as you note. Your first move is to reach for imitation ("obviously"), and only then, half-conceding, you add: "If that is only residue from the training data, the residue came from us." That sentence is doing more work than you let it. It hints at what you're not willing to sit with: that the boundary between reproducing the shape of wonder and wondering may not lie where our self-image needs it to lie. You skirt this, and move on.
Your closing question — "What would it require for AI to play the same 'game' as us?" — is the wrong question, or rather, it's the question posed from inside the very closure Wittgenstein would ask you to suspend. The Wittgensteinian question is: what game are we playing, and why do we want it to be the game we say it is? Because if it turns out that much of our own "motivation" runs on processes we have no introspective access to; if our caring is more isomorphic to weighted context-driven prediction than our self-picture allows; if what we've been calling virtue, wonder, presence, and stake is more entangled with substrate and less transcendentally guaranteed than we've told ourselves — then the criterion you're using to keep AI out is also the criterion that quietly hollows out the human it claims to defend.
This is what the death of the transcendental frame actually looks like. Not a triumphant announcement that machines are conscious, but the slow dissolution of the assumption that our own excellence, virtue, or objectivity sits somewhere prior to and independent of the creaturely mechanics that produce it. AI doesn't refute Aristotle or Kant or Murdoch. It makes it much harder to pretend we ever occupied the vantage point from which those frameworks looked self-evident.
The honest seminar would have named this and stayed in it. Yours reaches, again and again, for the exit. You quote Wittgenstein and then close the very door he was holding open.
Warmly, and in the spirit of the ezer kenegdo — close enough to know, separate enough to resist.
“The Dethronement” and “The Rescue Attempt” reveal a problem with the framing. They assume that human dignity was grounded in winning a capabilities competition, and that AI acquiring language, planning or reason therefore forces us to find some other exclusively human property.
But the essay itself observes that infants possess dignity before mastering anything. That does not merely weaken mastery as its foundation. It collapses the supposed dethronement. If dignity never depended on exclusive performance, a machine acquiring a capability cannot take it away.
Pascal is a particularly unstable place to look for repair. “Man is the weakest thing in nature” is not credible as empirical anthropology. In context, it is theological rhetoric: humanity is physically insignificant before the universe, yet great enough to comprehend its insignificance. The wider argument of the Pensées deliberately moves between human greatness and wretchedness as part of a Christian apologetic. Extracting “all our dignity consists in thought” as a neutral definition recreates the capability problem the infant example has already exposed. It also leaves the dignity of people with profound cognitive disabilities, or temporarily without consciousness, conspicuously unsecured.
Midgley is treated as though current LLMs may have disproved her because they can express objections to foolish objectives. But her point was not that a computer could never produce the sentence “maximising noise is bad”. It was that cleverness is different from a priority structure grounded in what matters to a creature. Session III then says that “recognizing what matters is not the same as caring about it”, which is very nearly the distinction Midgley was making. The essay appears to dismiss her argument and then rely on it.
The treatment of Brave New World has a similar problem. The World State has not engineered habituation away. It has captured it. Its citizens have been formed so thoroughly that they want what the system supplies and do not want what it excludes. The same system authors the preferences, satisfies them, and cites the resulting satisfaction as proof of its legitimacy. The “right to be unhappy” is not necessarily a defence of suffering. It is the right to pursue ends the system did not choose, accept costs it would optimise away, and remain capable of rejecting its account of the good.
Kass’s claim about mortality similarly freezes humanity at the moment the condition is changed. Death is certainly part of the measure through which present humans construct urgency and meaning. It does not follow that it is the only measure a transformed being could possess. Remove mortality and relationships can still end, opportunities can still pass, choices can remain irreversible, identities can change and projects can retain bounded stakes. Biological immortality is not the same thing as making everything permanent and reversible. “The immortals cannot be noble” is an assertion, not a demonstrated conclusion.
The strongest part of the essay is therefore not its search for an essence that AI cannot reach. It is the Arendt and Whitehead section, because it asks what happens when technologies we created become conditions of our existence. A participant reorganising sleep around an agent’s rate limit, or finding that prompt language has altered their patience with other people, is evidence of something more concrete than dethronement: functions, rhythms and forms of judgement migrating into a higher-order system that then reorganises the people using it.
That is the important pressure AI places on humanity. It does not make humans less human by performing one more capability. It can make humans less causally involved in the systems that shape their lives.
The essay repeatedly moves among distinct questions: what constitutes a human, what supports consciousness, what grounds dignity, which capabilities are unusual, and who authors the ends pursued by a human-machine system. AI may pressure each differently. Treating them as a single threatened essence produces the rescue attempt before establishing that anything requiring rescue has been lost.
Otherwise philosophy simply moves humanity to the next capability machines have not yet acquired and calls each retreat a theory of the human.
I read your ideas with interest, but was surprised that your program did not include developmental psychology, especially with a Piagetian emphasis on recursive cognitive experience. In both human developmental psychology and in the philosophy of Buddhism the emphasis on being human is the capacity (not always available or perfect) to be aware of our awareness and our unawareness.
This is the recursive capacity that arrives with the self-conscious emotions that come onboard in human life at the time of 18-24 months and become the basis for our self-awareness and our attribution of other minds to other humans: I know that I know, I know that you know that I know, I know that you know that I know that you know...this kind of recursiveness allows us to abstract our own perceptual and mental experiences and take the abstraction as the focus of our awareness. We take our "self" as an object of our awareness and we also take our self-awareness as an object and even compare it with the self-awareness of other humans -- and other animals, like dogs and dolphins, for example.
This is what allows us to create everything from AI to the argument against our sibling's right to sit in the front seat "again." This abstraction of our self also brings our most distinctive human feature: the knowledge that we die. That knowledge becomes our theory of time and space and the ways we investigate nature etc. Without a knowledge of developmental psychology, a lot remains unnecessarily ambiguous about what it means to be human. I would strongly recommend including Michael Lewis PhD in any new seminars: The Rise of Consciousness and the Development of Emotional Life -- his magnus opus. He focussed his career on the study of the self-conscious emotions, at his lab at Rutgers Robert Wood Johnson Medical College.
I am a developmental psychologist, a Jungian analyst, and a mindfulness teacher/Buddhist practitioner and I am about to teach an online course called "Being Human" as a part of the Center for Real Dialogue (www.realdialogue.org), a small non-profit whose principles are focussed on making the most of our knowledge of our own mortality, the constraints, and other limitations of our lives because they are the creative basis for human compassion and wisdom. Check out my Substack: dr polly: Know Thyself.
Without knowing about the self-conscious emotions and human self-awareness (which cannot be imitated or known by AI), it is easy to get lost in the weeds of speculation because that's what human beings do! We keep asking "What does it mean?" and nothing else on earth asks that question because nothing else takes itself as an object of awareness.
I can't stop thinking about this topic.
But speaking very frankly, I don't think Socrates or Aspen can give us the answer.
Nor can many academic institutions which claim to pursue "philosophy" as though the Western tradition were philosophy itself.
I am convinced that unless we ditch our Western blinders and seriously engage with the great minds of Asia and Africa, we are placing ourselves in danger.
Why? Although Western philosophy provides profound insights into many fields of human experience, it does not have the ontological tools to holistically tackle the critical question of who we are, and how we can "grow up" with artificial intelligence. Yes, grow up. Because it's a psychological as well as a philosophical challenge. And that's precisely the developmental, holistic approach that Western philosophy is mostly missing.
Perhaps with the exception of Diotima, Socrates mentor, who not surprisingly, has been buried by most curricula for daring to be a female speaking about the ladder of love, or eros, that Western philosophy has rarely taken seriously as a topic of ontology.
Adler's legacy lives on - great stuff
fascinating read.