# I have a big challenge for you. I want you to turn this draft, which I've been working on for mon... ## Retrieval Notes - Session id: `b469116f-471a-4aed-8707-7edcd77c706c` - Last activity: `2026-06-10T11:57:47.031Z` ## My Notes <!-- Add your notes here. This section is preserved across syncs. --> ## Conversation ### User I have a big challenge for you. I want you to turn this draft, which I've been working on for months—years even, into a fully polished philosophical text using what I'm giving you here as the core. Because this is an experiment, I'm giving you quite a lot of liberty to do what you want. But please note that I don't want a shallow piece of work. There are a lot of details in here, and I don't want you to give me a less rich version for the sake of making your life easier. I want this rich, dense, substantial piece of philosophy. Sections zero, one, two, and three—assume that what you can read there, even if it's not structurally perfect or philosophically perfect yet, is the argument and the general shape and the ideas I want to cover for each of those sections. Section four, on the other hand, is much rougher, much looser. Basically, you can see the beginning of the idea here, but it's nowhere near finished at this point. And remember, this is not an easy job. You need to look at my publications to see: A, the sort of way I write at a very fine grain; B, how I structure sections and arguments; C, where I add detail and where I pull back. I almost guarantee you that what you give me, I'm going to say it's not—like how I would write a paper. So see if you can surprise me by making it so I don't have to complain about that once you're done. Another thing you should mention is you can use any and all resources available to you, either in my vault or in my learning folder, or you can download more resources online if you would like. Although I doubt you'll need to be doing much, getting many more resources than we have right now. So yeah: Philosophically rich, structrually elegant. that's what i want /Users/nickyoung/Agent Vault/projects/generating-philosophy.md I just realized you should probably check this note as well. That should give you some very up-to-date information on what I want from this paper. Philosophically, I mean. # 0. Introduction The last decade or so has seen the rise of generative artificial intelligence: systems that produce text, images, code, music, video, and other outputs in response to prompts. AI has had success in domains where the value of an output is not exhausted by its superficial fluency. For example, in February 2026, researchers working on gluon scattering amplitudes gave GPT-5.2 worked examples for three, four, five, and six particles and asked it to find the general formula. GPT-5.2 proposed a closed-form expression; another internal model supplied a proof; and the authors then verified the result. The resulting paper argues that single-minus tree-level gluon amplitudes, often presumed to vanish, are non-vanishing in certain half-collinear configurations (Guevara et al. 2026). There are also recent examples in mathematics (Novikov et al. 2025), biomedicine (Gottweis et al. 2025), and materials science (Zeni et al. 2025). In this paper we argue that we should expect similar success in philosophy. %%this needs to be replaced with the recent maths discovery%% Specifically, we argue that current-generation LLMs are capable of producing philosophical texts that are _worth reading_. This phrase might seem loose, but that is part of its point%%not how i write%%. We do not want to begin by settling what counts as _good_ philosophy. Instead, we appeal to a distinction that anyone reading this text will recognise. You have read texts that are worth reading, and you have read texts that are not. As you begin reading this article, you likely hope that it is worth reading, in the sense that the time spent reading it will not be wasted. When you write a philosophical text yourself you aim to make it worth readers' while to read it, and whether or not the journal you send it to accepts it, depends on whether or not they agree. Two clarifications are needed. First, a text’s being worth reading is not the same as its being correct. A text can repay attention even if one rejects its conclusion: it may sharpen a distinction or answer an objection in a way that changes the dialectical situation. Second, the minimal unit we are concerned with is not the bare conclusion of an argument, but the argument itself. If an LLM output consists only in a pronouncement on some philosophical topic ('Direct Realism is correct', 'We should be utilitarians'), it is hard to see why it would be worth reading in and of itself, for the same reason that a bare pronouncement by a human philosopher would not be worth reading.[^1] The next three sections develop the main argument. Section I rejects the challenge from authorship: the claim that an LLM output cannot be philosophy worth reading because no philosopher lies behind it. Section II turns to abduction and argues that the absence of human-style inference to the best explanation in the producer does not preclude abductive structure in the product. Section III considers phenomenology and argues that the lack of consciousness does not prevent LLMs from producing philosophy grounded in phenomenology. --- # 1. The Challenge from Authorship In this section we address what we might call the _challenge from authorship_: the idea that philosophy is something that only persons, or at least minds, can produce. This view has not, to our knowledge, been explicitly defended in just this form, but it gives shape to an intuition that many philosophers may have: philosophy is a person-only domain. An imperfect comparison is with art. One might deny that an image generated by an AI system, at least in the familiar prompt-and-output cases, is an artwork because no artist exercises the relevant kind of intentional control over its production. One might think, for similar reasons, that philosophy can only be done by people. No text produced by an LLM can be a work of philosophy, because no philosopher lies behind it. Consider also that, like art, the study of philosophy is often focussed on individuals. Philosophy undergraduates take a course on Kant's ethics, or Lewis' metaphysics, and even at more advanced levels one finds specialists, conferences etc. spotlighting the work of specific philosophers. Compare this to the sciences: as a rule, scientific ideas, theories, discoveries etc. are the focus, not the individuals behind them: one does not find scientists who specialise in the work of Newton, or of Einstein; nor do biology departments teach Crick's view of DNA rather than Watson's. %%is this paragraph accurate re: science?%% We will try now and make this intuition more precise by continuing the comparison with artworks and philosophical works. We shall do this by considering the degree to which Davies' *performance* theory of art can be transposed to philosophy. He writes: > The work — what the artist achieves — is the process eventuating in that product. Works themselves are neither structures nor objects simpliciter, nor are they contextualized structures or objects [...]They are, rather, intentionally guided generative performances that eventuate in contextualized structures or objects. (p. 98) On Davies’ view, when a painter paints a picture, the canvas is what we attend to, but it is not the work. The work is the artist’s intentionally guided activity in producing that canvas; the canvas is, in Davies’ terms, the "focus of our appreciative interest in the work" (2004, p. 151). This is why provenance matters to him in a deeper way than it would matter on a view that identifies the artwork with a product plus contextual properties.%%unclear%% Facts about how the object came into being help determine what the work is and what is properly appreciated in it. If the same model were transposed to philosophy, an LLM text would fail not because it is badly argued, but because the relevant kind of philosophical performance is missing. Consider what is involved in attending to a Vermeer. We are not only registering a coloured surface%%not how i write%%. We are taking that surface as the outcome of a certain painter’s activity, in a certain historical context, with certain resources and limitations. Davies presses this point through cases in which perceptual sameness, or near-sameness, fails to settle artistic identity or appreciation.%%not how i write%% A canvas might emerge by accident from a washing machine and happen to look like a Rembrandt (Danto 1981); in that case, there is a Rembrandt-like surface, but no artistic performance of the relevant kind. Or a canvas might be presented as a Vermeer when it was in fact painted by van Meegeren[^1]; in that case, there is an artistic performance, but not the one the work was taken to make available. The point is not just that provenance gives us extra information. It is that provenance can change what we take the work to be and what kind of achievement we take ourselves to be appreciating. If Davies is right, the surface does not by itself settle the work. Here is the analogous proposal for philosophy. A philosophical text is not itself the philosophical work. The text is the product of the thinking, writing, and philosophising done by a person or group of persons over time. The text is therefore the focus of our attention, but only as a way of accessing the philosophical performance that brought it into being. On this proposal, a philosophical work is not identical with the sequence of sentences on the page. The text is the product of someone’s activity of thinking through a problem and giving that activity argumentative form. Reading the text is then a way of engaging with that activity. The authorship challenge is therefore not just a worry about missing biography. It is the stronger claim that, if no one has done the relevant philosophising, there is no philosophical work to which the text gives access. The question is whether this transposition should be accepted. We do not think it should. Davies has a reason to move from product to performance in the case of art: production history can affect which work we are dealing with and what is available for appreciation. A Rembrandt-like surface produced by accident is not a Rembrandt~~; a van Meegeren presented as a Vermeer is not the work it is taken to be~~. The philosophical case is different.%%not how i write%% If two texts contain the same argument, including the same inferential moves, the same considerations count for and against them. Their philosophical merit does not vary with the route by which the words came to be written. When we assess a philosophical paper, we ask whether the text does philosophical work. ~~Does it introduce a distinction that helps? Does it answer an objection that would otherwise remain pressing?~~ These questions do not require us to look behind the text to the philosopher’s activity. The grounds for the judgement lie in the argument as presented, not in the history of its production. This is where the analogy with Davies breaks down%%stupid way of putting things%%. Two papers that read identically do not differ in argumentative merit: they make the same moves and face the same objections. In the art case, production history can change what the work is. In the philosophy case, it changes, at most, what we think about the producer or the process by which the text came about. The organisation of analytic philosophy reflects this. Journals often strip author information from submissions before sending them to referees, and they do so because facts about authorship are treated as possible sources of distortion. The point is not that blind review always succeeds, or that philosophical practice is never interested in authors. The point is narrower: in this central evaluative context, the paper is supposed to be assessed by attending to what it says, not by reconstructing the circumstances under which it was written. A point from Dellsén et al. (2024) helps to articulate the same thought, although their concern is philosophical progress rather than LLM authorship. On their view, philosophical progress is "for-whom" rather than "by-whom": it consists in putting people in a position to increase their understanding, usually by making philosophical ideas publicly available (2024, p. 679). For present purposes, the useful thought is that philosophy makes its contribution through public materials that others can take up: arguments, theories, distinctions, thought experiments, and ways of framing problems. If this is correct, we should be cautious about locating the philosophical work behind the public text, in the process by which the text came about. The public text is not a dispensable trace of philosophy; it is where the philosophical contribution becomes fully available.%%this paragraph could be clearer%% The challenge from authorship is therefore a constitutive challenge. It treats the philosopher’s activity not merely as something that causes a philosophical work to exist, but as part of what the work is. On this picture, even a text indiscernible from a philosophical paper would not be philosophy if no philosophical activity lay behind it. We have argued that this should be rejected. If a novel philosophical text were produced by the wind blowing sand into a readable pattern, or by a very faulty washing machine, that would not, in and of itself, prevent the resulting text from being worth reading.%%some of this seems a bit redundant%% What remains%%unclear%% are not objections about what philosophy is, but objections about whether LLMs can produce texts with the relevant philosophical properties. While the authorship challenge argued that text produced by an LLM cannot be philosophy worth reading in virtue of the fact that it was produced by an LLM, these *capacity* challenges, on the other hand, can be thought of as claims that LLMs in their current state cannot produce philosophy worth reading, because LLMs lack features that are required to write worthwhile philosophy. %%maybe add an analogy with animals/young children –if they could write philosophy then it would 'count' as philosophy, but they do not have capacities such as language that seem necessary to be able to do philosophy%% In the next section, we consider the challenge from abduction, %%extremely succinct description of next section%%Section 3 turns to the parallel concern that some philosophical texts require phenomenal materials available only to conscious subjects. ### Footnotes 1. reference the ai image literature here, and mention that in most cases it is hard to imagine images being created without a human influencing things at least in some way. ↩ 2. Note that such a view does not amount to the denial that LLMs can produce beautiful images. We shall return to this point later. ↩ [^1]: mention who van meegeren was. --- # 2. The challenge from abduction # Section 2 — The Challenge from Abduction (distilled draft) Even if one accepts our argument that LLMs should not be ruled out automatically from producing worthwhile philosophy one might still think that such systems, at least in their current form, lack particular capacities which are needed to produce philosophy worth reading. In the next section we shall consider whether LLMs' lack of phenomenology impedes their ability to produce worthwhile philosophy. Before that we shall examine the charge that LLMs cannot perform abductive inference. Abduction, or inference to the best explanation, is reasoning from a body of evidence to the hypothesis that would best explain it. Abduction differs from deduction in that the evidence does not settle which explanation is correct. In a deductive argument the premises fix the conclusion: if all men are mortal and Socrates is a man, then Socrates is mortal, and there is no wriggle room. Now, imagine walking into your kitchen and finding the floor wet. What has happened? The wet floor does not determine the answer in the way the two premises gave you Socrates' mortality: a burst pipe would have left the floor wet, and so would a spilled bucket. But, given that the window is open, the water is under the window, and it rained last night, rain coming through the window seems the most plausible answer. To reason in this way, that is, deciding what best explains a set of facts, is common in everyday life and in the sciences alike. A scientist chooses one theory over another when it explains the same data more simply: Copernicus's model of the solar system was preferred to Ptolemy's because it explained the observed planetary motions without the elaborate epicycles the older model required. Williamson holds that philosophy is continuous with the sciences in its method: a philosophical theory, like a scientific one, earns acceptance by explaining the relevant data better than its rivals — more fully, and more simply — rather than by being proved, since deductive argument only passes the question back to its premises, which must themselves be chosen somehow (2007; 2021, p. 351). The same methodology has been defended for metaphysics in particular, where rival theories are weighed by these explanatory virtues (Sider 2011; Paul 2012), and it suits a conception of the discipline's aim that Sellars made famous: to understand 'how things in the broadest possible sense of the term hang together in the broadest possible sense of the term' (1962). We shall assume, in what follows, that producing philosophy worth reading depends, in large part, on abduction.[^1] Floridi and colleagues hold that large language models do not perform abductive inference. They describe what such models do instead as zeroth-order abduction: > LLMs seem to perform a kind of zeroth-order abduction: given a prompt, they generate a plausible continuation (a hypothesis or explanation) based purely on learned associations. In reality, their operation is driven by maximising the probability of the sequence... The model does not understand what an explanation is, but it produces text that follows the typical phrasing and structure of explanations. It does not reason about causes from scratch but outputs typical causes for typical effects observed in the training data. (Floridi et al. 2025, p. 9) An LLM, on this account, can produce a plausible hypothesis for a given body of observations, and, when the candidate explanations are explicitly provided, it can select the most suitable of them 'remarkably well, often at a near-human level' (Floridi et al. 2025, p. %%add page%%, citing Bhagavatula et al. 2020 and Balepur et al. 2024). What it cannot do is test a hypothesis. Inference here divides in Reichenbach's way: abduction supplies a candidate in the context of discovery, the candidate is then assessed against new data in the context of justification, and LLMs perform only the first part (2025, p. %%add page%%). The model's words connect only to other words, never to the things the words are about — the predicament Harnad (1990) named the symbol grounding problem — so an output may be the simplest and most unifying account of the data on offer and still be false, with nothing in the model able to find out. The proper use of such systems is then as brainstorming machines, churning out candidate hypotheses: 'It then becomes the human's task to carry out the justification phase' (2025, p. %%add page%%). The case can be made stronger than its authors make it. A sizeable literature has since set about measuring abduction in language models, and its results appear to confirm the diagnosis: models remain markedly weaker on abductive benchmarks than on deductive ones; where a long mystery story supplies the evidence and the task is to identify the culprit, they fail far more often than human readers; where every hypothesis consistent with a set of facts must be produced rather than recognised among options, performance collapses; and wherever generation can be compared with selection, generation lags (Salimi et al. 2026).[^2] On this reading the record shows exactly what the mechanism story predicts — the conceded part of abduction, choosing among candidates somebody else supplies, intact, and the denied part missing. So strengthened, the challenge holds that LLMs lack the capacity for abduction, and that the lack shows up wherever it is measured. The tasks on which the models fail have a common shape. In each there is a fact of the matter laid down in advance — a culprit, a diagnosis, a missing premise — and the task is to recover it; an answer counts as correct when it matches that hidden fact, and as nothing otherwise. Whether philosophical abduction asks for anything of this kind depends on what makes one explanation better than another, and 'best', in Lipton's treatment, can be read in two ways: the best explanation may be the likeliest, the most warranted by the evidence, or the loveliest, the one that would, if correct, provide the most understanding — 'Likeliness speaks of truth; loveliness of potential understanding' (2004, p. 59). The readings come apart: that opium puts people to sleep because of its dormitive power is, in Lipton's example, about as likely as an explanation gets, since it commits to almost nothing, and for the same reason it explains nothing (2004, pp. 59–60). A benchmark that scores the recovery of a concealed fact is scoring likeliness, with the world, or the puzzle's setter, holding the answer. When a philosophical theory is preferred because it would explain the relevant data better than its rivals, and more simply, what is judged is loveliness, and there is no answer sheet against which the judgement could be checked. Nothing in this standard requires the explanation to be true. Newtonian mechanics remains as lovely an explanation of the old data as it ever was, however unlikely the evidence for relativity has since made it (Lipton 2004, p. 60), and philosophy keeps its own counterpart: the positions a student is set to read contradict one another, so most of them are false, and their being false has never been treated as a reason to stop reading them — which is the position of an output that satisfies every explanatory criterion and is nevertheless false. Nor is a verdict pending that would change this: Copernicus's hypothesis had observations still to come, and a philosophical theory is not waiting on anything of the kind. The knowledge it must explain is knowledge already gained — by the sciences, by common sense, by philosophy itself — articulated and on the table when the theory is proposed (Williamson 2021, p. 356 %%check page%%), and the justification phase, such as it is, consists in argument over that shared stock: an opponent points to something the theory cannot accommodate, or produces a rival that explains it better, much as mathematics settles its first principles abductively and without experiment (Williamson 2021, pp. 356–7 %%check pages%%; Pigliucci %%match §3 citation%%).[^4] What the standard does demand can be read in the text that tries to meet it. Suppose a text argues that rain through the open window, rather than a burst pipe, best explains the wet kitchen floor, on the ground that the water lies under the window. That ground discriminates: the two hypotheses differ over where the water should be, so citing its location favours one and counts against the other. A text that instead offered the ground that the floor is wet would fail — both hypotheses lead us to expect a wet floor — while exhibiting, sentence for sentence, the typical phrasing and structure of an explanatory comparison. To this extent the challenge is right: explanatory structure can be had without any discrimination between rivals. The capacity in dispute is the capacity to produce texts of the first kind, reliably and not by accident. In the kitchen case anyone can tell that the location of the water discriminates and that the wetness of the floor does not; between philosophical theories the same judgement calls for training, and not because the rules are taught. Lipton is frank that 'the weakness of our grasp on what makes one explanation lovelier than another is discouraging' (2004, p. 61): there are no mechanical rules that generate a unique hypothesis from given data (2004, p. 83), and what counts as a lovely explanation is fixed in part by earlier explanations that serve as exemplars, and by prevailing styles of reasoning (2004, p. 139). The competence is, as Lipton himself observes, like our competence with grammaticality: systematic, reliable judgement, governed by exemplars rather than by statable rules (2004, pp. 6–7).[^3] Whether a regularity of this kind — unstated, graded, carried in examples — can be picked up from text is the question on which the challenge now turns. The exemplars that carry the standard are themselves text. Earlier explanations and styles of reasoning survive in one form, writing, and the corpus on which an LLM is trained contains that writing: the literatures of the sciences and of philosophy, together with the far larger record of ordinary argument, weak as well as strong. However natural it is to think that a system which only continues text could not pick up a judgement from it, regularities of just this kind are what such systems demonstrably acquire. A model is never given the rules of English grammar — no complete statement of them exists to give — yet its text is grammatical; no theory specifies which strings of English are meaningful, and its output stays meaningful all the same (Wolfram 2023): regularities for which nobody possesses the rules are installed by fitting the text that contains them. The challenge itself traces the model's explanatory appearance to its training on human explanatory practice, so the position needed is that grammar and meaningfulness can be acquired from text while the standards of explanation cannot, although all three are regularities of the same kind, unstated, graded, and carried in examples; nothing in the mechanism story supplies a reason for the difference. What steers abductive theory choice is, on Williamson's own description, an aesthetic sense 'surely connected to a capacity for abstract pattern recognition' (2021, p. 369 %%check page%%), and there is reason, then, to expect a model trained on this record to produce text whose explanations are good, by the same route by which its text comes to be grammatical and meaningful. Where these systems fail is just as regular. A small transformer trained on sequences of matched brackets learns the language for ordinary cases and breaks down where success requires explicit counting, while networks of the same kind, shown a smudged digit, settle what it is at a glance, with no criterion they could state (Wolfram 2023). The line falls between graded judgement on open-textured material, which fitting to examples yields, and exact, step-by-step bookkeeping, which it does not; weighing rival explanations belongs on the first side, the long exact derivation on the second. And the record assembled above breaks along the same line: the tasks at or near human level ask for comparative judgement over supplied or short material, and the tasks that collapse ask for exhaustive enumeration, or for the recovery of a single concealed fact, scored right or wrong against it. Read with Lipton's distinction in hand, the record confirms this placement rather than the challenge: what fails is abduction with an answer sheet, which philosophy does not ask for, and what succeeds is comparative judgement of the kind it does. None of this is at odds with the description of the mechanism: a true description of a process at one level does not displace true descriptions at another, as Lipton notes of the parallel deflation of explanationism by Bayesianism — like arguing that thinking about technique cannot help one's squash game because the ball's motion is governed by mechanics (2004, p. 108). And what is acquired is a capacity, exercised well or badly on particular occasions: an output can carry the apparatus of a comparison whose cited difference favours neither side, and which kind of text it is can only be settled by reading it, as for every philosophical text. Half of the challenge has not yet been touched: what was conceded was selection, and in the record generation lags it wherever the two can be compared. The denial of unaided generation has independent support — systematic theorising, on Williamson's picture, requires new distinctions at a more abstract level not given in the data (2021, p. 350 %%check page%%) — and a sharp current statement: what LLMs lack, Zahavy argues, is 'the generation of novel explanatory hypotheses' (2026, p. 1), partly because generating hypotheses begins in sense experience, a claim about the materials of theorising that the next section takes up, and partly because a system cannot validate the novel candidates it produces, which can be answered here. Producing candidates only from what one has already encountered disqualifies nobody: we rank only the potential explanations that have occurred to us, and the actual explanation is sometimes one that nobody has thought of (Williamson 2021, p. 358 %%check page%%) — why suppose that any of the explanations we happen to have thought of is true (Lipton 2004, p. 70)? Inquiry, for everyone, is conducted from inside whatever stock of candidates history has supplied. Nor is producing them a lottery: against Hempel's conclusion that hypotheses are 'happy guesses', Lipton argues that most hypotheses consistent with the data are non-starters, and that the contrastive structure of the evidence — a fact with a foil — sharply constrains which candidates could explain it at all (Hempel 1966, p. 15; Lipton 2004, pp. 82–3). Generation is guided by the same considerations as selection, and learned from the same exemplars. A genuinely new distinction, finally, is new in relation to the literature: new when the literature lacked it, whoever first set it down, and a line drawn where none had been marked is not a recombination of the lines already there. Even the discipline's transformations mostly began as moves of recognisable kinds — apparatus imported from a neighbouring field, an assumption everyone had treated as binding called into question — and became transformations through what the discipline went on to do with them. The demand that remains, that a producer be able to tell which of its own novel candidates are apt, asks for an answer sheet that exists for no one: the written record preserves questionings that succeeded and questionings that failed, verificationism as thoroughly as rigid designation, with nothing in the producing of them marking the one kind off from the other. Which succeed is settled by the discipline's subsequent work, and a model's candidates enter that process on the same terms as anyone's; whose entries they are — the model's, or the prompter's — is the question of Section 4. To sum up, the challenge asked of philosophy a kind of abduction it does not practise. What the measurements show failing is abduction graded against an answer sheet, and philosophy grades its explanations against none; what it does demand — graded comparative judgement, carried in exemplars, exercised in the reading — is a regularity of the kind such systems acquire from text, so there is reason to expect the capacity, though no guarantee of any particular output. What experience contributes to philosophy's materials comes next. [^1]: Not everyone accepts that the explanatory virtues carry the same weight in philosophy as in the sciences: see Bueno and Shalkowski (2020) and Thomasson (2015). [^2]: As of mid-2026: selection between two supplied explanations runs close to human accuracy (roughly 87–88 against a human 91.4); long narrative mysteries remain well below human readers (42.9 against 47); short, tightly constrained formal tasks approach ceiling (99.6) while collapsing where every admissible hypothesis must be produced (21.5); and ranked diagnosis degrades sharply when candidates are not supplied (Salimi et al. 2026 %%check figures against their Table 3%%). The survey organises the field by Lipton's two stages of hypothesis generation and hypothesis selection. [^3]: Loveliness inherits what Lipton calls Hungerford's objection: beauty is in the eye of the beholder, and explanatory loveliness may be too subjective and interest-relative to do epistemic work (2004, p. 70). His reply is that judgements of loveliness are stable enough across competent practitioners to ground appraisal (2004, pp. 141–4), and nothing stronger is needed here, where the standard is used to assess texts rather than to underwrite a theory of inductive warrant. [^4]: Philosophy does increasingly draw on experimental results, and Williamson's own examples include the psychology of perception; but the experiments are executed by the relevant scientists, and their results enter philosophical argument as published findings (2021, p. 356 %%check page%%) — that is, in articulated form. --- # 3. The challenge from phenomenology A further capacity worry concerns phenomenology. Few would say that LLMs are conscious, and we will assume the same here; yet this might seem to pose a problem for LLM philosophy, or at least for philosophy grounded in, or making use of, phenomenology. Some philosophy interrogates or refers to what it is like to see red (Harman, 1990), to feel anger (Goldie, 2000), or to have a particular intuition take hold (Chudnoff, 2011). If LLMs lack conscious experience, it seems as if this might hamper their ability to produce worthwhile philosophy which relies on it. This is not to say that all philosophy would be off bounds: large stretches of philosophy of language and modal metaphysics proceed without leaning on the phenomenology of any particular experience.%%not how i write%% %%to abrupt%%Zahavy’s discussion of a thought experiment of Einstein's brings out this worry: > Einstein’s variation required inventing new axioms based on a physical intuition that did not yet exist in the mathematics. He envisioned a physicist inside an elevator being uniformly accelerated through deep space. Inside this enclosure, the sensory experience reveals a specific pattern: when objects are released, the floor rushes up to meet them. To the physicist, the objects appear to fall with identical acceleration, regardless of composition. Thus, the simulation here was not a permutation of symbols, but a manipulation of perceptual experience. (Zahavy 2026, §5) The thinker imagines[^4] some set of circumstances and attends to what would be experienced within it — in Einstein’s case, that all objects inside the elevator would appear to fall with identical acceleration. That observation becomes the new axiom: a starting point arrived at through experiential simulation rather than formal derivation, from which further reasoning proceeds. If thinking of this kind depends on simulated experience, then it would seem to be out of reach for LLMs. They can provide descriptions of weightlessness or elevators, but they have never felt the sensation of an elevator descending, let alone weightlessness.[^2] Philosophy also uses experience based thought experiments. Jackson’s Mary case turns on what it is like to see colour, and we might think that as with Einstein's thought experiment, it provides us with an experiential axiom, from which further philosophical reasoning can proceed. The same worry then arises in philosophy: experience based thought experiments seem to require what LLMs do not have.[^3] %%abrupt –needs to signpost the difference between science and philosophy%% Pigliucci offers an account of philosophy on which it is constrained by, but does not aim at, the world as the natural sciences do. He writes: > This means that the basic parameters that philosophers use as their inputs, the starting points of their philosophizing, their equivalent of axioms in mathematics and assumptions in logic (or rules in chess) are empirical data about the world. This data comes from both everyday experience [...] and of course increasingly from the world of science itself. (Pigliucci, p. 6) Philosophy begins from worldly materials, but those materials function as starting points for conceptual exploration. Pigliucci elaborates this picture by drawing on Smolin’s account of evocation, taking chess as the paradigm. Positing the rules of a game does not require that they pre-exist; once posited, they generate a structure with rigid properties — a space of consequences that can be explored but not chosen. Once the rules of chess are codified, all the facts about chess become demonstrable, even though chess did not exist before its rules were written down. Pigliucci’s claim is that philosophy operates in this register. Unlike the rules of chess or the axioms of mathematics, however, the starting points of philosophising are constrained empirically. They are constrained by how the world actually is, including by what experience is like. This is why Pigliucci distinguishes philosophy from fiction: philosophy is not merely the invention of imaginary possibilities. As he puts it: > Philosophy, I maintain, is in the business of doing empirically informed evoking, not inventing. (Pigliucci, p. 7) %%This term is used in a technical sense%% The same picture covers philosophical thought experiments%%not how i write%%. Even when philosophers explore possible worlds or imagined scenarios, they do so “with an interest in figuring things out as far as this world is concerned” (Pigliucci, p. 7). The thought experiment articulates an axiom — an experiential or empirical starting point — and the philosophical work proceeds within the conceptual landscape that axiom evokes. This brings out a difference between the elevator and Mary cases. Both are evocations of the kind Pigliucci describes: each posits an experiential axiom and develops what follows from it. What differs is what the evocation is for. In Einstein’s case, the evoked structure yields a hypothesis whose status is then settled by experiment — the elevator gave him the equivalence principle, but the principle’s truth was a matter for empirical confirmation. In Mary’s case, the evoked landscape is itself the object of inquiry; the philosophical question is what the landscape contains, not whether anything outside it corresponds. The role of the evocation is what tracks the disciplinary difference. %%not how i write%%Evocation is present in both cases%%elaborate%%; what differs is whether the evoked structure is the means to an external test or is itself the object of inquiry. %%too abrupt. make sure that people understand what articulated phenomenology%% No competent discussant of the knowledge argument has personally undergone her transition. Once the case is articulated, work on it is work on the articulation. Responses to Jackson press at the level of the articulated structure, not at the level of any discussant’s experience. Lewis’s reply, for instance, modifies what is taken to follow from Mary’s situation, not what Mary’s situation is taken to be like from the inside. What allows the Mary case to do philosophical work in public is its articulation: the experiential material it draws on has been made available in language. This is the form in which phenomenology enters philosophy generally. The articulation is what does the philosophical work; the experience the articulation refers to need not be undergone by the people working on it. Philosophers work on the experiences of the blind and on the experiences of non-human animals without first-hand access to either, by working on the articulations the literature has accumulated. The point matters for LLMs in a particular way. They have no raw phenomenology of their own; but no text corpus contains raw phenomenology either. What a corpus contains is articulated phenomenology, and it is in articulated form that phenomenology becomes usable in philosophical argument. %%check this paragrpah it has been rearranged%% Merleau-Ponty’s discussion of self-touch raises a sharper question — that of phenomenological _discovery_. When one fingertip touches another, one finger plays the role of toucher and the other of touched. The roles can reverse, but not simultaneously: at any given instant, the body is split between touching and touched. Suppose the toucher-touched asymmetry was first identified by Merleau-Ponty himself, by sustained attention to his own embodied experience. The asymmetry would then be a phenomenological axiom out of reach of any LLM not trained on Merleau-Ponty or his interlocutors: an axiom an LLM could not have produced for itself, because the system lacks the body and the experience that the discovery requires. But that does not prevent an LLM from working philosophically on the description once articulated. What survives, then, is a narrower asymmetry. Even granting that LLMs can work within articulated landscapes %%not good%%, some phenomenological articulations seem to be originated through first-person attention; LLMs have no experience to attend to. First-person attention is one route to an articulation; it is not what gives an articulation philosophical use. What makes an articulation philosophically usable, on Pigliucci’s picture, is not its causal origin but its functioning as an axiom — its capacity to evoke a landscape with rigid properties %%not good%%. %%this sentence is important and better than surrounding sentences.%%An articulation can also be arrived at by working from the articulations a corpus already contains, generating new ones by extension and recombination. %%enrico hates this sentence, too flowery too pompous%%Whether a candidate articulation succeeds is a question about what it evokes, and that question is answered the way other philosophical questions are — by the public assessment of the conceptual structure the articulation makes available. It is the assessment any candidate articulation, whatever its origin, must finally meet. The phenomenology objection rests on a producer-to-product inference: that the absence of experience in the producer must remove phenomenological value from the product. The inference fails. LLMs lack conscious experience, but phenomenology enters philosophy as articulated content. Pigliucci’s account explains why this is not a workaround. Philosophy uses empirical and experiential materials by turning them into constrained spaces %%too much jargon%% for conceptual exploration. Since those spaces are public and inferentially usable once articulated, current models can produce phenomenology-based philosophy worth reading. [^2]: footnote saying that he calls it manipulative abduction. it should probably also explain why we might think go this as abduction as well as what we talked about in the previous section [^3]: A nice example in the footnote will be the feeling of understanding that is sometimes used as a way of motivating cognitive phenomenology. --- # 4. The Challenge from Authorship 2 # Section 4 The Challenge from Tools –*this section will definitely begin with some non bullet point form of pretty much exactly this text.* - In sections two and three, we argued that, despite being unable to make inductive inferences or possessing phenomenology, there is still good reason to think that LLMs can generate text that possesses these properties.%%Should try to avoid the repeat of argued.%% - In this section we shall consider another, final, challenge. Stated bluntly the challenge is: if philosophy worth reading is produced by a model, this is the work of the prompter, not the model. LLMs cannot produce philosophy worth reading in the same sense that a typewriter cannot, but both are tools which can be *used by* a philosopher to produce philosophy worth reading. - At a certain fineness of grain, this is trivially true. Consider a philosopher who puts a section of a worthwhile paper into an LLM and tells it to produce one without any spelling errors or typos. If the LLM performs this task correctly, then in a certain sense we might think that it has produced worthwhile philosophy. The charge here however is that to believe an LLM responsible for the valuable properties of a philosophical text is akin to believing that it is the ventriloquist's dummy which is doing the talking. - This line of attack is bolstered when we consider the sorts of answers that LLMs give when asked philosophical questions. Asking 'philosophical questions' to a chatbot (e.g. what is the correct philosophical theory of consciousness? What is the meaning of life?), will be met with a bland survey of possible positions at best and turgid, content-free 'slop' at worst. The fact that the dummy only speaks when the ventriloquist is holding it, makes clear who is really actually talking. - - The fact that the only people with a chance of getting better answers that this are philosophers, this is all the more reason to think that they are the authors of what LLMs output. > [!danger] Claude: Nothing Here Is Settled > Nothing in this section is settled. Do not assume that I want Section 4 to be anything like what is here at the moment — neither content-wise nor structurally. These are working notes, not commitments. Treat everything below as provisional raw material. - whenever an LLM text displays the properties of worthwhile philosophy, such properties are the work of the prompter, not the model. - - not having phenomenological experiences and making inductive inferences, there is still good reason to think LLMs should produce text that possesses these properties. - The intuition to grant first: when a philosopher gets worthwhile philosophy out of a model, the model is the thing they did it with. The philosophy is the philosopher's. A word processor earns no credit for the paper composed on it, and a model earns no credit for the philosophy composed with it. - This is the same intuition the brush invites in the Midjourney case. We do not credit the brush with the painting; we credit the painter, who used it. - The support for treating the model as an instrument: left to itself, asked a philosophical question directly, a model returns a bland survey. It runs on without producing anything worth reading, the way a typewriter left running would produce nothing. Good results require a person directing the process, which is what one would expect if the model contributed nothing of its own. - So the absence of worthwhile philosophy that is the model's, and the poverty of unaided output, are one claim with its evidence: the model is an instrument, and unaided output is what an instrument does when no one guides it. ## 4.2 Where the intuition is right - Unaided output often is a bland survey. - Worthwhile output is often closely directed by a person. - The person who writes the prompt is the author of the prompt. - A typewriter and a word processor do earn no credit for what is written with them. ## 4.3 The move the challenge makes - From: the human wrote the prompt and directed the process. - To: the philosophy is the human's. - The step from a fact about the producer to a conclusion about the philosophy is the step Section 1 refused. What the producer did does not settle what the text contains or whose the contents are. - The challenge needs more than the bare tool intuition to license the step. It needs the model to be a tool in the specific way a typewriter is: a thing that adds nothing to the content and fixes only what the user has already settled. ## 4.4 If the model is a tool, it is not a tool like a typewriter - This is the Midjourney move. The brush is a tool, but pressed, it is unlike other tools. The same pressure applies here. - A typewriter adds nothing to the content of the novel. It fixes in type what the author has settled. Every word was the author's before the machine touched it. - The model adds to the content. What a prompt supplies and what the output contains come apart, and the next subsections say how. ## 4.5 The prompt is a starting point, not a body of philosophy - A prompt posits a starting point. In Pigliucci's sense, the starting point evokes a landscape: a structure that did not exist before the positing and that, once posited, has properties no one chooses. - Codifying the rules of chess is the model for this. The person who writes the rules authors the rules. The theorems of chess follow from the rules and are not chosen by whoever wrote them. - Writing a prompt is writing rules of this kind. The consequences of the starting point are no more the prompt-writer's than the theorems of chess are the rule-writer's. ## 4.6 The output develops consequences the prompt does not contain - For the typewriter description to hold, the philosophy in the output must already be in the prompt, so that the model relays it. - A starting point, once posited, has more consequences than anyone has drawn, and they hold whether or not anyone draws them. - The output can develop a consequence the prompt does not contain and that could not be read off the prompt. - So the philosophy in the output is not in the prompt. The relay description fails, and with it the typewriter description. ## 4.7 The philosophy is the model's, not no one's - One retreat remains: grant that the philosophy is not the prompt-writer's, and say it is no one's. The consequences follow from the landscape on their own, so the model produced nothing. - The landscape makes the consequences available. It does not state them. - Stating them is producing a text that develops them, and the model does this. - Sections 2–3 license the step: the model produces a text with the philosophical properties without the producer-side act a human would perform. It develops the landscape without standing in the relation to it a human enquirer stands in. - What is worth reading is the developed text. The developed text is the model's. ## 4.8 Why unaided output is not a counterexample - The survey output develops no posited starting point. Nothing has been evoked for it to work out. - An instrument left running produces nothing because no starting point has been set, not because the model can produce nothing. - So the poverty of unaided output supports the claim rather than telling against it. It shows that the worthwhile cases are the ones where a starting point was posited and worked out, which is where the model does the developing. ## 4.9 Iterative use - A person often works with the model in turns: reading an output, redirecting, cutting, asking for development in one direction. - The challenge says that in working this way the person is doing the philosophy, so the philosophy is at least partly the person's. - Each turn the person takes is a fresh starting point or a narrowing of the one in play. Choosing which line to pursue is choosing where to develop, not developing. - The person's work divides into positing starting points and assessing what the model returns. - Positing is the chess-rule-writer's contribution. - Assessing is the editor's or the referee's contribution. - An editor who picks out good papers, and a referee who recognises a good argument, exercise philosophical judgement without authoring what they pick out or recognise. Section 1's blind-review observation returns: the assessor attends to what the text does and is not its author. - As the interventions grow finer, assessment shades towards co-writing. Where it does, the contribution is shared. Even then the person selects among developments the model produced, and the developments are the model's. ## 4.10 What the paper's claim requires - The claim is that LLMs can produce philosophy worth reading. One clear case suffices. - The clearest case is the one where a person posits a starting point and the model develops it, with little fine-grained intervention. - Heavy-collaboration cases can be granted as shared authorship without loss to the claim. - Undecided: how much of the collaborative range to claim as the model's. ## 4.11 Pressure points - A rich prompt. - A detailed prompt fixes a great deal, so the development is mostly contained in it. - A detailed prompt is a larger starting point, not a worked-out philosophy. Developing its consequences is distinct from it, as a longer axiom set is still distinct from its theorems. Detail increases what is posited; it does not place the development inside the positing. - The landscape does the work, not the model. - The consequences follow from the starting point on their own, so the model only reports them. - The consequences follow from the landscape; the text developing them is the model's. A consequence being available is not the same as its being stated. What is read is the statement. - Selection as authorship: treated at 4.9. ## 4.12 Closing position - The model is a tool, but not a tool like a typewriter. A typewriter adds nothing to the content; the model develops consequences the prompt does not contain. - Writing the prompt, the properties of the evoked landscape, and the text that develops them are three things. The first is the person's; the second is no one's; the third is the model's. - This says why a worked-out output is the model's, and why unaided survey output is not a counterexample to the claim. ## ~~Section 4 — Moves (revised)~~ - ~~If philosophical evaluation concerns intrinsic virtues of texts — elegance, unity, non-ad-hocness, combining simplicity with strength — then the question of whether LLMs can produce good philosophy is the question of whether they can produce texts exhibiting these properties. Sections 1–3 established this framing and argued that process-based objections do not undermine it. What remains is the constructive case: can LLMs actually produce such texts, and if so, how?~~ - ~~I want to grant Floridi et al.'s diagnosis completely. LLMs are "engines of generative plausibility": "given a prompt, they generate a plausible continuation (a hypothesis or explanation) based purely on learned associations. In reality, their operation is driven by maximising the probability of the sequence" (Floridi et al. 2024). They perform "zeroth-order abduction" — producing outputs that exhibit explanatory structure without selecting those outputs by comparing alternatives. All of this is correct at the level of mechanism. But statistical probability is relative to training data. What the model has learned to treat as "plausible" depends entirely on what it was trained on. So the question becomes: what does the training data encode?~~ - ~~The philosophical corpus is not a random sample of text. It is the output of a multi-level filtering process that selects, at each stage, for properties tracking Williamson's intrinsic virtues.~~ - ~~Peer review selects for handling of objections, engagement with the literature, non-trivial contribution — filtering out the arbitrary and ad hoc.~~ - ~~Citation selects for arguments that prove useful — arguments other philosophers find themselves needing to address, refine, or build upon — filtering for explanatory power and integration with existing work.~~ - ~~Teaching and anthologising select for clarity, illumination, and pedagogical power — filtering for elegance and unity.~~ - ~~Sustained philosophical attention selects for depth — works that reward re-reading because their arguments have structure worth unpacking.~~ - ~~The filtering is noisy: bad philosophy gets published, popular but mediocre work gets cited more than excellent but obscure work. But noisy filtering is still filtering. The tendency is toward virtue, even if individual data points deviate.~~ - ~~This claim requires empirical grounding — the proportion of academic philosophy in training data, the actual degree of filtering, and the training pipeline's selection mechanisms are questions that should not be answered by stipulation. What follows assumes that the tendency exists and is non-trivial, not that the filtering is perfect or comprehensive.~~ - ~~An LLM trained on this corpus learns the distribution of text that has survived these filters. The learned probability distribution is shaped by the intrinsic virtues — not because the model has been instructed in those virtues, but because texts exhibiting them are overrepresented in the training data relative to texts that lack them. Williamson writes: "Apart from its relation to E, the more T has the intrinsic virtues of a good theory, the better (ceteris paribus). It should be elegant and unified, not arbitrary, gerrymandered, ad hoc, or messily complicated. It should be informative and general. In brief, it should combine simplicity with strength" (2024, p. 354). The filtering process selects for exactly these properties. The virtues are therefore _latent_ in the model: implicit in the statistical regularities of the learned distribution, recoverable from the model's outputs, but not explicitly represented as rules or criteria the model applies.~~ - ~~This is like the relationship between a language model and grammar. A model trained on grammatical text produces grammatical outputs without having been taught grammar as a set of rules. The grammatical patterns are latent in the distribution — the model has absorbed them from the data without being given the rules explicitly. Similarly, a model trained on philosophically filtered text produces outputs tending toward philosophical quality without having been taught the evaluative criteria. The quality patterns are latent in the distribution. This is not a claim that every LLM output is good philosophy, any more than every output is grammatical. It is a claim about the tendency of the distribution — the direction in which the probability landscape slopes.~~ - ~~Even Zahavy concedes the relevant competence. He grants that LLMs can handle deductive work from given materials and explicitly restricts his critique: "we emphasize that this proposal is specifically tailored to the physical sciences, where the object of study is external material reality. In abstract domains such as Mathematics or Computer Science, the Sense Experience (E) may be grounded in high-dimensional topology or have other goals such as generality or minimality" (Zahavy 2026). Philosophy is one of those abstract domains. Its materials — arguments, distinctions, thought experiments, the logical space of positions — are textually available. They are the training data. The E→A jump that Zahavy claims LLMs cannot make is a jump from bodily experience to formal axioms; in philosophy, the "axioms" are already articulated in language and already in the corpus. Williamson himself notes that philosophy's evidence base includes "whatever knowledge the natural and social sciences, philosophy, and common sense have already gained" (2024, p. 356) — and this knowledge is textual.~~ - ~~But latent does not mean automatically expressed. Unprompted, LLMs produce generic, hedging text — surveys, overviews, cautious summaries. The intrinsic virtues are in the distribution but are not the default output. If they were, every LLM response on a philosophical topic would be good philosophy, which is manifestly false. A model trained on virtue-filtered text can produce texts exhibiting those virtues, but the capacity is not exercised by default. The prompt determines when it is.~~ - ~~The prompt determines which region of the continuation space the model generates from. The probability distribution the model has learned extends over a vast space of possible continuations. The prompt constrains which region the model generates in. Different prompts access different regions, and these regions differ in how reliably they exhibit intrinsic virtues. A bare question — "What is consciousness?" — activates a region dominated by survey-type text: cautious, generic, low in philosophical quality. This is the most probable continuation because it is the most common type of text following such prompts in the corpus. A dialectically structured prompt — one that lays out a position, identifies its vulnerability, and gestures toward a repair — activates a different region, where the most probable continuation is a philosophical _move_: the next step in the dialectic.~~ - ~~The prompter's skill consists in writing text whose good continuation — in the statistical sense of "most probable given the learned distribution" — is also good philosophy. Three modes of prompting access increasingly virtue-dense regions of the distribution:~~ - ~~Dialectical framing (one-shot, problem-oriented): pose a question embedded in dialectical context — not "what is X?" but "given these considerations, what follows?" or "the obvious objection is Y; address it." The training data is densely populated with such dialectical responses at the appropriate points in the argumentative structure. Walton, Reed, and Macagno's argumentation schemes formalise this: each scheme comes with licensed "critical questions" — the canonical pressure points. These are exactly the moves the corpus contains thousands of instances of, and exactly the moves a well-prompted model will produce.~~ - ~~Solution-gestured prompting (one-shot, solution-oriented): write a paragraph that points toward a solution without fully articulating it, so the good continuation is the next step in developing that solution. Richer than dialectical framing because the prompt itself contains philosophical content — it begins an argument, and the model continues in the direction indicated.~~ - ~~Conversational iteration (multi-turn): the prompter and the model produce philosophy together in an iterative process — write, continue, refine, develop, object, repair. Each turn further constrains the continuation space. The intrinsic virtues of the emerging argument increase with each round because each round further specifies what "good continuation" means. This mode sits on a continuum of autonomy: the prompter provides direction, constraints, and editorial judgment; the model provides dialectical moves, articulation, and pattern-completion. Neither is doing philosophy alone; what they produce together is a text exhibiting intrinsic virtues.~~ - ~~In a corpus filtered by intrinsic virtues, what Floridi calls "plausible continuation" and what Williamson calls "exhibiting intrinsic virtues" are not independent properties. They are correlated — because the filtering shaped what counts as plausible. The discipline produced text; the filtering selected text exhibiting intrinsic virtues; the filtered text became the training data; the LLM learned the distribution of the filtered text; the LLM's "plausible continuation," in the right context, therefore tends to exhibit the intrinsic virtues encoded in the distribution. This does not require the LLM to understand the intrinsic virtues, or to apply them as criteria, or to evaluate its outputs against them. It requires only that the training data was shaped by those virtues — which it was, because that is what philosophical filtering consists in. Floridi et al. themselves raise the question: "if an AI can generate the same explanatory hypothesis a human would, does it matter that the process was different? From an epistemological standpoint, perhaps yes — justification is significant — but regarding the content of the hypothesis and our interpretation of it, maybe not" (2024). For philosophy, the answer to their question is: it does not.~~ - ~~Lipton's distinction between likeliness and loveliness illuminates why this convergence holds. The "likeliest" explanation is the most probable; the "loveliest" is the one that "would, if correct, be the most explanatory or provide the most understanding" (Lipton 2004, p. 59). These can diverge: a conspiracy theory may be lovely (it unifies many apparently unrelated events) without being likely. But in a corpus filtered for loveliness — where the texts that survived peer review, citation, and anthologising are those judged illuminating, elegant, and explanatorily powerful — the likeliest continuation in the model's learned distribution tends also to be the loveliest in Lipton's evaluative sense. The filtering has aligned statistical probability with philosophical quality. Williamson further notes that "we rank only those potential explanations that have been thought of" (2024, p. 355). The philosophical corpus is the record of what has been thought of — and what survived the filtering. The model has absorbed this ranked space.~~ - ~~The "just statistics" dismissal confuses levels of description. Lipton: "arguing that Inference to the Best Explanation is wrong because Bayesianism is right is like arguing that thinking about technique cannot help my squash game because the motion of the ball is governed by the laws of mechanics" (2004, p. 108). The ball obeys mechanics whether or not you think about technique, but the mechanical description does not make the technique description idle. Similarly, an LLM's outputs are generated by stochastic processes over token distributions — and those outputs exhibit philosophical structure: they handle objections, draw distinctions, illuminate subject matter. The stochastic description and the philosophical description operate at different levels. Both are true. The fact that the mechanism is statistical does not settle the question of whether the outputs meet philosophical standards, because philosophical standards concern the output, not the mechanism.~~ - ~~The obvious worry: if the LLM is producing continuations shaped by existing filtered text, can it produce anything genuinely new? Williamson notes that "enumerative induction is inadequate for systematic philosophical theorizing, which often requires introducing new distinctions at a more abstract level not given in the data" (2024, p. 353) — and gives Dummett's distinction between assertoric content and ingredient sense as an example of a conceptual innovation that "cannot simply be read off the data." The model has learned not just particular arguments but patterns of argumentative _structure_ — patterns of how distinctions are drawn, how arguments are constructed, how positions are developed. These structural patterns can be instantiated in novel ways, producing arguments that do not appear verbatim in the training data but follow the patterns the training data established. Most philosophical innovation — most published, cited, taught philosophy — consists in exactly this kind of reconfiguration at higher levels of abstraction. The rare framework-introducing genius may be beyond current LLMs. But the bulk of what the discipline values does not require that kind of genius. And novelty, while not itself an intrinsic virtue on Williamson's list, is implicit in the virtues he does list: a theory that merely restates what is already known scores low on informativeness and generality — two of the virtues a good theory must have.~~ - ~~Two empirical questions arise. First, how much can a general-distribution LLM — one trained on the full breadth of human text, not specialised for philosophy — produce texts exhibiting intrinsic virtues? Second, would specialist training on philosophical texts improve performance? If the first question receives a positive answer and the second adds comparatively little, this suggests something about what philosophy is. Sellars characterised philosophy as the discipline concerned with "how things in the broadest possible sense of the term hang together in the broadest possible sense of the term" (_Philosophy and the Scientific Image of Man_, 1962). A system trained on the full breadth of human knowledge — on science, history, literature, law, ordinary discourse — has, in a sense, been trained on precisely the subject matter Sellars identifies as philosophy's own. The striving toward general intelligence, even if unachievable within current architectures, may itself be what positions these models for philosophical work — not because they have been taught philosophy specifically, but because they have absorbed the broadest possible range of how things hang together. If this is right, it deepens the encoding claim: the intrinsic virtues may be latent in the model not only because the philosophical corpus is filtered for quality, but because the general corpus encodes the breadth of connection that philosophical argument draws upon.~~ - ~~The paper itself is an instance of the process it describes. If the reader judges its arguments clear, its distinctions illuminating, its engagement with objections substantive, then the paper exhibits the intrinsic virtues it discusses — and these virtues are partly the product of the human-LLM collaboration it argues for. The paper was produced with a general-purpose LLM, not a system specialised for philosophy — which is itself evidence bearing on the two questions just raised. The paper does not need to demonstrate LLM philosophy as a separate exercise. It is a demonstration, submitted for blind review, evaluated by the very criteria it articulates.~~ - ~~Humanity asked a computer to do philosophy. It received the answer '42' — correct, according to the machine, but meaningless to the questioners, because they had never known what the question was. The problem was not with Deep Thought's capacities but with humanity's prompt. The intrinsic virtues were latent in the machine; what was missing was the right question to draw them out. Now we know what the question is — and we know that the answer, when the question is well-formed, can exhibit the philosophical qualities that the discipline has spent centuries learning to value.~~ --- # References # References Frankish, K. (2024). What are large language models doing? In A. Strasser (Ed.), *Anna's AI Anthology: How to live with smart machines?* (pp. 55–78). Xenomoi. ### Assistant I'm taking this on as a full deep-writing-mode job: ground in your project note and published prose, extract the sources the paper leans on, then produce the polished text. Starting with the project note and the skill's grounding references. ### Assistant Now the remaining references, plus locating your published papers and the project's source folder. ### User [Request interrupted by user]