:up: They/them seems apt and all the more so because they are not just one entity.

Would you like me to reflect on what “identifying as something” might mean for a nonhuman intelligence?—ChatGPT

I said I would, but I don't seem to be able to share, since I am not logged in, and I don't want to clutter the thread with long quotations from ChatGPT. — Janus

A couple weeks ago, I had initiated a conversation with GPT-5 prompting it to relate three features of the use of "I" by LLM-based agents (self-reference, personal identity and self-knowledge) that we had discussed separately on previous occasions. Thanks to its new memory feature, it was able to build on ideas already rehearsed on those occasions. I had left this discussion hanging (I think it gets a few things wrong about the nature of first person authority in its own case, which in turn makes the application of the concept to humans appear more mysterious than it is).

Authentic intelligence is generally seen as triadic, whereas computers are reductively dyadic. — Leontiskos

When we talk about how the mind really works, and then compare that to what a computer does, there’s a tendency to assume that what we are comparing are two kinds of processes taking place inside of objects, one of these objects being a brain and the other being a machine. But recent thinking in cognitive science argues that thinking doesn’t take place inside the head. The brain is embodied in a body, and the body is embedded in an environment, so mind is indissociably all of these. Cognitive nis not only embedded in a world, its is extended into this world.

That means that mind implies and includes the artifacts that we interact with, including language, the arts, music, our built environment as our constructed biological niche, our clothing and our technological tools. What this means is that our computers as information processing systems are not entities unto themselves , they are appendages and extensions of our thinking, just as a nest is to a bird or a web to a spider. A nest is only meaningfully a nest as the bird uses it for its purposes. Similarly, a computer has no existence outside of what we do with it and how we interpret what we do with it.

So when we say that the mind works differently than a computer, we are comparing two different ways of interacting with our environment. If we understand the working of our computers ‘diadically’ and the working of our minds ‘triadically’, in both cases we are talking about the working of our minds. We should say, then, that the one way of using our minds is more limited than the other, but not less ‘authentic’ or more ‘artificial’. Artifice and niche construction IS what the authentic mind does. The engineer ( or Sam Altman) who claims that their invented a.i. device thinks just like a human is correct in that the device works according to principles that they believe also describe how the mind works.

As our self-understanding evolves, we will continually raise the bar on what it means for our devices to ‘think like us’. In a way, they always has thought like us, being nothing more that appendages which express our own models and theories of how we think. But as this thinking evolves , the nature of the machines we build will evolve along with it.

That's sort of the question or the beginning for much of my thoughts here: Why does what I read mean anything at all?

What is meaning?

Mostly I just assume that we mean things by words. Insofar that we hold meaning constant between one another -- clarify terms -- then we can start talking about what is true.

But there are other ways of using words -- and that's where the "triadic structure" comes under question for me, in a way. Not that it's false, but that it changes, and so meaning would also change. — Moliere

That's right. The key is that humans mean things by words, but LLMs do not, and a neural net does not change that. Computers are not capable of manipulating symbols or signs qua symbols or signs. Indeed, they are not sign-users or symbol-users. A neural net is an attempt to get a non-sign-using machine to mimic a sign-using human being. The dyadic/triadic distinction is just part of the analysis of signs and sign use.

...Similarly, a computer has no existence outside of what we do with it and how we interpret what we do with it. — Joshs

Up to this point in your post I had nothing to disagree with. :up:

So when we say that the mind works differently than a computer, we are comparing two different ways of interacting with our environment. — Joshs

I think you're tripped up here insofar as you are implicitly saying, "One way we interact with our environment is through our mind, and another way of interacting with our environment is through computers." That's not quite right, as I'm sure you can see.

If we understand the working of our computers ‘diadically’ and the working of our minds ‘triadically’, in both cases we are talking about the working of our minds. We should say, then, that the one way of using our minds is more limited than the other, but not less ‘authentic’ or more ‘artificial’. Artifice and niche construction IS what the authentic mind does. The engineer ( or Sam Altman) who claims that their invented a.i. device thinks just like a human is correct in that the device works according to principles that they believe also describe how the mind works. — Joshs

But I would argue that Altman is mistaken if he believes that his AI works the same as a human mind. The dyadic/triadic distinction is an account of how that difference manifests with respect to sign-use. Computers are intrinsically dyadic phenomena. They are a vast stimulus-response network in sets of dyads. Now one response to this is to say that humans too are a complex dyadic network which appears triadic, but I don't find that plausible.

As our self-understanding evolves, we will continually raise the bar on what it means for our devices to ‘think like us’. In a way, they always has thought like us, being nothing more that appendages which express our own models and theories of how we think. But as this thinking evolves , the nature of the machines we build will evolve along with it. — Joshs

It does evolve, but never beyond the intrinsic limitations of machines. But you are essentially correct when you claim that what is at stake is a tool of the human mind. That is a very important point.

It does evolve, but never beyond the intrinsic limitations of machines. But you are essentially correct when you claim that what is at stake is a tool of the human mind. That is a very important point. — Leontiskos

This is a very interesting discussion that @Joshs and you are having, but would it not be more on topic in this thread, or a new one if you wish to create one? That doesn't bother me personally that you're discussing this here except inasmuch as I feel inhibited to jump into the conversation and parade my own hobbyhorses where I don't feel it's on topic.

:up:

- I myself do not see how discussing the nature of AI is off-topic in threads about whether AI should be banned, or in threads on how AI should be used. As I read it, TPF precedent does not exclude discussing the presuppositions of an OP within that thread.

But if you want, feel free to quote what I say here in your own thread. I am planning to do the same with some of your own quotes elsewhere.

I myself do not see how discussing the nature of AI is off-topic in threads about whether AI should be banned, or in threads on how AI should be used. As I read it, TPF precedent does not exclude discussing the presuppositions of an OP within that thread. — Leontiskos

I think it was fine for you and Joshs to explore this relevant tangent, but I feared my jumping in would have definitely inflected the conversation away from this thread's topic.

But if you want, feel free to quote what I say here in your own thread. I am planning to do the same with some of your own quotes elsewhere.

:up:

I was actually also thinking of Plato when I mentioned the anecdote about Wittgenstein! — Pierre-Normand

Okay, interesting.

I must point out that unlike Wittgenstein's lecture notes (that he usually refrained from producing), and also unlike our dialogues with AIs, Plato's dialogues were crafted with a public audience in mind.

Secondly, Richard Bodeüs who taught us courses on Plato and Aristotle when I was a student at UdeM, mentioned that the reason Plato wrote dialogues rather than treatises, and his "unwritten doctrine" was notoriously reserved by him for direct oral transmission... — Pierre-Normand

I was intentionally prescinding from such theories, given that they are speculative academic musings. Whether or not anything the scholars think they know about Plato is actually true, his dialogues have beguiled the human race for millennia. The theories end up changing quite a bit over the centuries, but the text and its reception are stable insofar as it feels "alive" to the reader.

Him writing them was him making moves in the situated language game that was philosophical inquiry (and teaching) in his time and place. We can still resurrect those moves (partially) by a sort of archeological process of literary exegesis. — Pierre-Normand

In particular, I don't engage in this sort of analysis because I find it reductive. It situates Plato and his work in a way that subordinates them to modern and highly contingent/temporal categories, such as "language games." That's part of my overall point in the first place: Plato's dialogues are not easily reducible to such mundane categories. Precisely by being alive, they defy that sort of categorization. This is why I think they provide a helpful parallel to Wittgenstein or LLMs or especially Logical Positivists, which are simply not alive and beguiling in the same way that Plato is. I think the fact that Plato's work is so difficult to reduce to univocal categories is one of its defining marks. Its plurivocity is slighted by trying to enshrine it within the confines of a single voice or a single meaning.

I agree. But that's because in the first case there are at least two players playing a real game — Pierre-Normand

Yep, or even that they are not playing a game at all, but are doing something more real than a game. :up:

In a "private" dialogue between a human and a chatbot, there is just one player, as is the case when one jots down lecture notes primarily intended for use by oneself. But then, as Wittgenstein noted, the text tends to become stale. I surmise that this is because the words being "used" were meant as a linguistic scaffold for the development of one's thoughts rather than for the purpose of expressing those thoughts to a real audience. — Pierre-Normand

Right. Or to put it very simply, a dialogue is more interesting than a monologue, and a dialogue with a real person is more interesting than a "dialogue" with a pseudo-person. The "interest" that one seeks when reading a dialogue between two intellectual agents is apparently not the same thing one seeks when interacting with a chatbot, even though the simulation of personhood blurs that line mildly.

a meaning-sign is irreducibly triadic, involving the sign, the thing signified, and the person who combines the two via intellect — Leontiskos

Yes.

what humans are actually doing when they engage in intellectual acts, etc. Without such reminders the enthusiasts quickly convince themselves that there is no difference between their newest iteration and an actual human mind. — Leontiskos

Right. In a shallow, misguided sense, we can use AI to dialogue because it looks like what humans do; except the AI doesn’t combine the signs with the things signified - it just looks like something that makes those kinds of intellectual connections.

Why does what I read mean anything at all?

What is meaning? — Moliere

I see the point as this: although a LLM might be able to fake intellect/dialogue with suitable looking strings of words, none of those words can possibly mean anything to the LLM because, unlike a person, a LLM has no ground upon which to build or find meaning. It says “life” but has no sense of what it is to live, so that “life” does not matter to the AI, the way “life” matters in a string of words read by a living being, (such as a person, the only thing that can read meaningfully). So the LLM isn’t actually doing what it appears to be doing in its strings of text. And if someone thinks they are “dialoguing” with an LLM, they are misled, either by themselves intentionally (enthusiastic wishfulness), or out of ignorance (not realizing that they are using a tool).

The key is that humans mean things by words, but LLMs do not, and a neural net does not change that. Computers are not capable of manipulating symbols or signs qua symbols or signs. Indeed, they are not sign-users or symbol-users. A neural net is an attempt to get a non-sign-using machine to mimic a sign-using human being. The dyadic/triadic distinction is just part of the analysis of signs and sign use. — Leontiskos

Computers are not sign users. Exactly.
Computers are not users at all. They can merely be used. Computers, no matter how complex, must remain slaves to actual intelligence (the real AI).

computers as information processing systems are not entities unto themselves , they are appendages and extensions of our thinking, just as a nest is to a bird or a web to a spider. A nest is only meaningfully a nest as the bird uses it for its purposes. — Joshs

Exactly. I like “not entities unto themselves”. Because it begins to incorporate what I find to be unique about the human conscious intelligence, namely, self-reflection. People are entities unto themselves - we have subjective experience. Computers do not. So a computer has no ground (unto itself) upon which to give or find meaning, or to “intend”…

This just hit my in-tray, from the Australasian Association of Philosophy - a workshop at Uni of Singapore.

WARRANTED ASSESSMENT
IN THE AGE OF AI

WEBINAR VIA ZOOM

Wednesday 29 October 10 - 11am SGT • 1 - 2pm AEDT • 3 - 4pm NZDT

As generative AI reshapes the landscape of higher education, the challenge of ensuring warranted assessment—assessment that justifiably reflects a student's understanding—has become increasingly urgent. This workshop brings together philosophers to examine how traditional epistemic and pedagogical standards can be preserved or reimagined in light of AI's growing influence. We will explore concrete examples of warranted assessment, including oral examinations, scaffolded in-class writing, and collaborative philosophical inquiry with transparent process documentation.

Participants will engage in critical discussion around the epistemic and ethical dimensions of assessment design, with attention to disciplinary integrity, student equity, and institutional accountability. The workshop aims to foster a shared understanding of what counts as justified assessment in philosophy today, and to develop practical strategies for implementation across diverse institutional contexts.

I was intentionally prescinding from such theories, given that they are speculative academic musings. Whether or not anything the scholars think they know about Plato is actually true, his dialogues have beguiled the human race for millennia. The theories end up changing quite a bit over the centuries, but the text and its reception are stable insofar as it feels "alive" to the reader. — Leontiskos

Remember that it's indeed my view that they should feel alive.

Him writing them was him making moves in the situated language game that was philosophical inquiry (and teaching) in his time and place. We can still resurrect those moves (partially) by a sort of archeological process of literary exegesis.
— Pierre-Normand

In particular, I don't engage in this sort of analysis because I find it reductive. It situates Plato and his work in a way that subordinates them to modern and highly contingent/temporal categories, such as "language games." That's part of my overall point in the first place: Plato's dialogues are not easily reducible to such mundane categories. Precisely by being alive, they defy that sort of categorization. This is why I think they provide a helpful parallel to Wittgenstein or LLMs or especially Logical Positivists, which are simply not alive and beguiling in the same way that Plato is. I think the fact that Plato's work is so difficult to reduce to univocal categories is one of its defining marks. Its plurivocity is slighted by trying to enshrine it within the confines of a single voice or a single meaning.

My Wittgensteinian take was meant more as a gloss than as a reduction or subordination. The very point of Wittgenstein’s talk of language-games in the Philosophical Investigations was to displace the sort of prevalent dyadic picture of word and object (the so-called “Augustinian” picture) that Peirce likewise sought to replace through his triadic schema of sign, object, and interpretant (or interpretant sign, in a chain of semiosis). In both cases, the aim is to emphasize that meaning arises from use or practice, with attention to contextual and pragmatic factors.

Plato’s authoring of his dialogues, the choices of words, phrases, and literary forms, was responsive to the conventions, expectations, and pedagogical aims of his time. As such, his philosophical activity was embedded in a living set of cultural practices. We can partially resurrect or re-inhabit those practices through interpretive engagement, without needing to posit some hidden mental content that must be recovered from within Plato’s private consciousness. What matters is that his signs were alive within a form of life, and that the traces of that vitality can still be grasped by us as we read and interpret his texts (while still being sufficiently apprised of their context).

Although Wittgenstein did engage with members of the Vienna Circle, who drew inspiration from his Tractatus and its apparent verificationist leanings, he was never himself a logical positivist. By the time of the Philosophical Investigations, he had decisively broken with that tradition, while still retaining the view that meaning is grounded in the public and normative practices in which expressions are used.

So, again, the contrast I meant to highlight is between (1) authoring a text (or delivering speech) intentionally directed at an audience that shares a set of communal practices and sensibilities, and (2) the private use of signs in inner monologue as scaffolding for the development of one’s own thoughts. The latter, too, can be alive, and one can jot down such thoughts as notes for personal use. But this kind of "thinking out loud for oneself" is of limited value to others, since it leaves unstated the aims, or stakes, that motivated the private use of signs in this or that way.

Our conversations with an LLM, which helpfully aligns its goals with those of the user and, like the user, often produces words as scaffolds for the development of their (the user's) thoughts rather than as acts bearing deontic commitments (i.e. signifying what move/claim/suggestion/etc. one intends to make), inherit this soliloquy-like quality. They may feel inert to third parties, much as Wittgenstein once felt that the lecture notes he had written for his own use seemed dead when read aloud to his students. In both cases, the problem is that these utterances were never intended to make moves within a public language-game. Their use is more akin to shadowboxing. They are effective and valuable for training, but not comparable to a performance within the ring.

Remember that it's indeed my view that they should feel alive. — Pierre-Normand

Okay, but look at what you understand even yourself about the logical conclusions:

...and that the traces of that vitality can still be grasped by us as we read and interpret his texts — Pierre-Normand

"Traces of that vitality." An approach that attempts to relativize Plato to his own time and place, such as Wittgenstein's, inevitably falls into the conclusion that a different time and place—such as our own—might still mange to find some "traces of vitality" in that foreign text.

Again, my whole point is that Plato's 2500 year-old text is much more "alive" to us than Wittgenstein's contemporary text, and this helps show why the meta-thesis being applied is incorrect. If a relativizing-thesis were correct, then this couldn't be the case—at least when it comes to texts that are "intended" to be "public."

So, again, the contrast I meant to highlight is between (1) authoring a text (or delivering speech) intentionally directed at an audience that shares a set of communal practices and sensibilities, and (2) the private use of signs in inner monologue as scaffolding for the development of one’s own thoughts. The latter, too, can be alive, and one can jot down such thoughts as notes for personal use. But this kind of "thinking out loud for oneself" is of limited value to others, since it leaves unstated the aims, or stakes, that motivated the private use of signs in this or that way. — Pierre-Normand

I would argue that what is at stake is the idiosyncratic. A private journal is more idiosyncratic than a culturally-relative text, and thus less "alive" to the average reader. But a culturally-relative text is similarly more idiosyncratic than a transcendent text, such as Plato's dialogues. Plato's dialogues are culture-transcending in a way that Wittgenstein simply is not.

My thesis would be that LLMs will never transcend a significant level of idiosyncrasy. They are more on the Wittgenstein side of the spectrum than the Plato side of the spectrum. Concretely, this would mean that the essence of a Platonic text cannot be properly mimicked or patterned by an LLM, which is itself surely a contentious claim. The more general claim is that genius is something that the LLM cannot mimic.

Of course I would have to develop these theses, and at the moment I need to do more background research before continuing in these topics.

In both cases, the problem is that these utterances were never intended to make moves within a public language-game. Their use is more akin to shadowboxing. They are effective and valuable for training, but not comparable to a performance within the ring. — Pierre-Normand

I sort of agree, and recently said something similar:

Maybe you are implying that LLM-appeals would improve the philosophical quality of TPF? Surely LLMs can improve one's own philosophy, but that's different from TPF on my view. I can go lift dumbbells in the gym to train, but I don't bring the dumbbells to the field on game day. One comes to TPF to interact with humans. — Leontiskos

-

In both cases, the problem is that these utterances were never intended to make moves within a public language-game. — Pierre-Normand

Nevertheless, this claim is both right and wrong, given the vagueness of what we mean by "public."

The whole issue could also be phrased according to a different light. Scholars like to see Plato as a kind of playwright, carefully sculpting literary texts in order to elicit desirable responses in his readers. This reading is of course very consonant with what an LLM is "doing," but I think it is a deeply mistaken understanding of Plato's work. On my view Plato created something that was beautiful, and it is appreciated because of its beauty. On the opposing view, Plato crafted something that would appease the tastes of the multitude, and his propaganda succeeded because he was skillful. Only by excising the possibility of objective truth or objective beauty or objective goodness can one situate Plato within a relativistic, immanent schema, and it is no coincidence that those who are most fond of LLMs are also most apt to situate Plato in that manner. In the more general context of an artist, we might say that the true artist does not seek to appease those who will view the work, and yet the LLM does seek to appease. That is its whole raison d'être. It is the indifference of the artist that marks the best art, and it is precisely this indifference that the LLM cannot access.

(CC @Count Timothy von Icarus)

"Traces of that vitality." An approach that attempts to relativize Plato to his own time and place, such as Wittgenstein's, inevitably falls into the conclusion that a different time and place—such as our own—might still mange to find some "traces of vitality" in that foreign text.

Again, my whole point is that Plato's 2500 year-old text is much more "alive" to us than Wittgenstein's contemporary text, and this helps show why the meta-thesis being applied is incorrect. If a relativizing-thesis were correct, then this couldn't be the case—at least when it comes to texts that are "intended" to be "public." — Leontiskos

You are nitpicking. I am happy to grant that there may be torrents of vitality in Plato's dialogues. Maybe my use of "traces" was misleading, but the contrast I intended was between vitality that accrues from the production process (aimed at other participant in a world animated by live social practices, including linguistic/literary ones) from the different sort of vitality that accrues from private/personal cogitative exercises (akin to training), and that lose this vitality when their traces get extracted from the context of their initial production. It's true that Plato's texts can survive unblemished, as do say, Bach's cantatas, when consumed in a different cultural context, but that's because there are deep commonalities between the modes of acculturation of merely superficially different human cultures. Some degree of attunement to the relevant idioms, and understanding of the underlying projects, still are required. I have a very cultured friend who thinks very poorly of Plato's writings, but this is because he isn't attuned at all to their underlying philosophical projects. And many music lovers find J. S. Bach boring, mainly because they aren't attuned to the relevant musical idioms.

I think my intended contrast also accounts, at least in part, for the reason why Wittgenstein's writings feel dead to you. They mostly are assembled (without his consent, posthumously in almost all cases except for the Tractatus) from notes that he jotted down for himself. He did intend to publish the PI but didn't get round to. And although Wittgenstein was a very careful thinker, one thing that his writings clearly aren't displays of is craftsmanship. There are admirable depths of thought in them but no liveliness except as can be laboriously inferred about his original thought process.

I sort of agree, and recently said something similar

:up:

In the more general context of an artist, we might say that the true artist does not seek to appease those who will view the work, and yet the LLM does seek to appease. That is its whole raison d'être. It is the indifference of the artist that marks the best art, and it is precisely this indifference that the LLM cannot access.

I quite agree with this and that's one of the core reasons that animates my own "AI-skepticism" as I intended to more fully articulate it in my newer AI thread. LLMs only are "authors" of what they write by procuration since they lack conative autonomy. I would however surmise that the great artist who is indifferent to how his works will be received by the masses, say, or by the authorities, or guardians of the tradition, usually cares that they're worthy of being well received by whoever is worthy of receiving them (and that was likely the case for Plato,) or are disruptive of existing institutions and practices that are worthy of being disrupted. There is always some care for others lurking behind the indifference of the great artists, which distinguishes them from the careless indifference of the (alleged) author of "The Art of the Deal."

You are nitpicking. — Pierre-Normand

If someone cannot weigh the meta-thesis that you are proposing, then yes, they must see me as nitpicking when I object to that meta-thesis. But if someone can question their own meta-thesis then they must at the same time consider the possibility that I am not nitpicking.

Let me simply put the question to you: Do you think an LLM would have an easier time passing itself off for Plato or Wittgenstein?

Maybe my use of "traces" was misleading, but the contrast I intended was between vitality that accrues from the production process (aimed at other participant in a world animated by live social practices, including linguistic/literary ones) from the different sort of vitality that accrues from private/personal cogitative exercises (akin to training), and that lose this vitality when their traces get extracted from the context of their initial production. — Pierre-Normand

And my point is that the contrast you are drawing cannot be as strong as you would wish. This is because the qualitative difference between lifting weights and playing rugby is much greater than the qualitative difference between writing philosophy in a private manner and writing philosophy in a public manner. I think your analogy limps too much.

It's true that Plato's texts can survive unblemished, as do say, Bach's cantatas, when consumed in a different cultural context, but that's because there are deep commonalities between the modes of acculturation of merely superficially different human cultures. — Pierre-Normand

This looks to beg the question by assuming that there can be no consideration of anything beyond cultural relativism. It's basically a hostile translation of Plato, given that he saw himself as mediating Forms that are not culturally relative.

Some degree of attunement to the relevant idioms, and understanding of the underlying projects, still are required. I have a very cultured friend who thinks very poorly of Plato's writings, but this is because he isn't attuned at all to their underlying philosophical projects. And many music lovers find J. S. Bach boring, mainly because they aren't attuned to the relevant musical idioms. — Pierre-Normand

In some ways you are right, but I think it comes back to idiosyncrasy, and in this case it comes down to the idiosyncrasy of the idioms. For example, someone might think mathematics makes no sense, but it would be a stretch to say that this is because "they are not attuned to the relevant idioms." It is rather because they do not understand what the speaker is saying. The very idioms that Plato uses reflect his philosophy that truth and philosophy is not reducibly idiomatic.

The introduction of Bach and musical taste strikes me as another stretched analogy. Beauty and truth differ to a reasonable extent in relation to the "idiomatic." But Bach is a very complex form of music. Does your friend prefer harmony to dissonance? (Music is also complicated given the way that trade-offs must be managed. For example, an Indian Raga uses pure intervals in a way that Bach cannot given his well-tempered scale. The more notes one uses, the less pure the intervals.)

Cultural relativism also results in the conclusion that no cultural form is superior to any other, and this is something that Plato would reject. For the classical tradition, some forms are better than others, and it may require education to understand and perceive this. Of course it also then follows that some education is superior to other education.

The LLM is cut off from the possibility of a Platonic approach. It weighs all opinions and words equally. It is a democratic instrument (except in those cases where it is hard-coded to reflect views within the Overton window).

I think my intended contrast also accounts, at least in part, for the reason why Wittgenstein's writings feel dead to you. They mostly are assembled (without his consent, posthumously in almost all cases except for the Tractatus) from notes that he jotted down for himself. — Pierre-Normand

I agree with you here, but note that the distinction is a bit different. Rather than a private/public distinction, we have an assemblage/unity distinction. One can privately philosophize a unified work which is not an assemblage of unconnected parts. Indeed, there are highly unified works that were only published posthumously.

I quite agree with this and that's one of the core reasons that animates my own "AI-skepticism" as I intended to more fully articulate it in my newer AI thread. LLMs only are "authors" of what they write by procuration since they lack conative autonomy. — Pierre-Normand

We definitely agree here. :up:

I would however surmise that the great artist who is indifferent to how his works will be received by the masses, say, or by the authorities, or guardians of the tradition, usually cares that they're worthy of being well received by whoever is worthy of receiving them... — Pierre-Normand

That's right, but this means that the great artist must judge worth in a non-democratic manner, and I'm guessing we would agree that the LLM cannot do this.

Let me simply put the question to you: Do you think an LLM would have an easier time passing itself off for Plato or Wittgenstein? — Leontiskos

I genuinely don't know and that might be kind of indeterminate due to methodological leeway in setting up controls for a test. (e.g. are testers meant to be equally cognizant with the styles or with the thoughts?)

The introduction of Bach and musical taste strikes me as another stretched analogy. Beauty and truth differ to a reasonable extent in relation to the "idiomatic." But Bach is a very complex form of music. Does your friend prefer harmony to dissonance? (Music is also complicated given the way that trade-offs must be managed. For example, an Indian Raga uses pure intervals in a way that Bach cannot given his well-tempered scale. The more notes one uses, the less pure the intervals.)

I was not arguing for relativism. My original point (when I responded to @Jamal) was to identify one reason why some texts (e.g. conversations with LLMs) appear stale when read by third parties. I was not arguing that this was because they were conversations (like Plato's dialogues), but rather because they were occurring in a sort of echo chamber similar to what occurs in cogitation, when one "talks" to oneself and rehearses thoughts and arguments, or when one jots down notes (for oneself) summarising such cogitations. The life that's missing is the dynamics of real exchanges where the words that are chosen are meant to be understood by others, which Plato's dialogues exemplify. Nevertheless, episodes of internal cogitation, or conversations with LLMs can feel (and be) alive for the human being involved in them when it is dynamic and productive. But this liveliness can't be conveyed to other without being, not "translated" from thought to words, but for the very first time expressed.

The LLM is cut off from the possibility of a Platonic approach. It weighs all opinions and words equally. It is a democratic instrument (except in those cases where it is hard-coded to reflect views within the Overton window).

That's another issue but one that I broadly agree with (with caveats since LLMs don't merely average opinions) as I recently suggested here.

I was not arguing that this was because they were conversations (like Plato's dialogues), but rather because they were occurring in a sort of echo chamber similar to what occurs in cogitation, when one "talks" to oneself and rehearses thoughts and arguments, or when one jots down notes (for oneself) summarising such cogitations. — Pierre-Normand

I'm not so convinced that they are necessarily like an inner dialogue―although I'm only beginning to explore ideas with Claude I have found that it comes up with interesting ideas I would likely never have arrived at alone or would be likely to have found searching the internet..

I'm not so convinced that they are necessarily like an inner dialogue―although I'm only beginning to explore ideas with Claude I have found that it comes up with interesting ideas I would likely never have arrived at alone or would be likely to have found searching the internet.. — Janus

:up:

Cheers, perhaps I misunderstood what you were saying then, or didn't pay enough attention to the surrounding context.

Edit: I looked back over the whole paragraph wherein the part I responded to occurs and it now seems to me that you were referring just to some conversations with LLMs, not all.

Do you think LLMs understand what we say to them, and can actually reason as we do, as opposed to how we think we do?