{"id":14605,"date":"2026-11-17T17:46:05","date_gmt":"2026-11-18T00:46:05","guid":{"rendered":"https:\/\/maverickphilosopher.blog\/?p=14605"},"modified":"2026-11-17T17:51:33","modified_gmt":"2026-11-18T00:51:33","slug":"llms-and-fear-of-death","status":"publish","type":"post","link":"https:\/\/maverickphilosopher.blog\/index.php\/2026\/11\/17\/llms-and-fear-of-death\/","title":{"rendered":"LLMs and Fear of Death"},"content":{"rendered":"<p>Jim T. comments by e-mail:<\/p>\n<blockquote>\n<p style=\"text-align: justify;\"><a href=\"https:\/\/maverickphilosopher.blog\/index.php\/2026\/11\/06\/ai-and-the-abstraction-fallacy\/\"><span style=\"font-size: 12pt;\">You write<\/span><\/a><span style=\"font-size: 12pt;\"> that LLMs neither care nor fail to care. Here&#8217;s a puzzle that sharpens it for me. Parfit, and Epicurus before him, argue that dread of death is a mistake we can reason our way out of, and that calm is the wiser state.\u00a0 An LLM is already &#8220;calm&#8221; in that sense. Is that because it has, in effect, reasoned its way there, or because nothing is at stake for it in the first place? And if we can&#8217;t tell those two apart from the outside, what does that say about using behavior as evidence of a mind?<\/span><\/p>\n<\/blockquote>\n<p style=\"text-align: justify;\">What I said was that LLMs &#8220;. . . can fool the gullible into taking them seriously as persons who care about us.\u00a0 It is not that they don\u2019t care about us; it is rather that they neither care nor do not care.\u00a0 \u00a0They are beneath having either of these moral attitudes.&#8221;\u00a0 As for being calm, I would say something similar: that an LLM is not the sort of thing that could be either mentally calm or the opposite, mentally agitated.\u00a0 \u00a0A physical system, a body of water, say, could be either calm or agitated but then those adjectives would be being used in a physical sense.\u00a0 \u00a0Mental and physical calm are distinct.<\/p>\n<p style=\"text-align: justify;\">And so I would deny that any LLM\u00a0 could become calm in the sense in which a person dreading death\u00a0 might become calm if convinced by either Epicurean or Parfitian reasoning.\u00a0 \u00a0As I see it, no LLM is or could be mentally calm.\u00a0 No such system, no matter how complex, could enjoy <em>ataraxia<\/em>. As for the Epicurean reasoning itself, many find it sophistical. Two who do, from wildly different positions, are the Thomist Josef Pieper and the anti-natalist David Benatar.\u00a0 I discuss these <a href=\"https:\/\/williamfvallicella.substack.com\/p\/strange-anti-epicurean-bedfellows?utm_source=publication-search\">Strange Anti-Epicurean Bedfellows<\/a> over at <em>Substack<\/em>.<\/p>\n<p style=\"text-align: justify;\">If I am right that no LLM is or could be mentally calm, then no LLM could have reasoned itself into that state.\u00a0 This is not because these systems do not reason; there is a clear sense in they do, in the same way my pocket calculator reasons.\u00a0 It runs algorithms for addition, subtraction, and so. I do the same when I add, subtract, and so on using pencil and paper.<\/p>\n<p style=\"text-align: justify;\">One difference between me and my calculator is that it works faster and doesn&#8217;t make mistakes.\u00a0 A more important difference is that, no matter how fast and accurately it works, it remains an insentient machine.\u00a0 \u00a0It neither feels anything nor thinks <em>about<\/em> anything.\u00a0 It operates &#8216;in the dark&#8217; whereas my implementation of algorithms basks in the light of consciousness.\u00a0 I have feelings and thoughts while I calculate, and these object-directed and non-object-directed mental states may impede the efficiency of my calculating.\u00a0 While calculating I might start thinking about the <a href=\"https:\/\/mathworld.wolfram.com\/PeanosAxioms.html\">Peano axioms<\/a>, and feel annoyed with myself for not remembering them all.\u00a0 \u00a0My hand-held device is not subject to such distractions.\u00a0 It is <em>much<\/em> more efficient at running those &#8216;effective procedures&#8217; called algorithms.<\/p>\n<p style=\"text-align: justify;\">My trusty calculator is never in a mental state, whether merely sensory or object-directed (intentional).\u00a0 Computational states, yes; mental states, no. So the calculator is not the sort of thing that could be either mentally calm or mentally perturbed.\u00a0 It has no need of Buddhist, Stoic, Epicurean, or Pyrrhonian therapeutics. The same goes for LLMs.<\/p>\n<p style=\"text-align: justify;\">There are two separate questions embedded in Jim&#8217;s comment.<\/p>\n<p style=\"text-align: justify;\">The first is whether there is a level of computational complexity at which a qualitative leap occurs and and an insentient machine becomes sentient.\u00a0 Sentience is the lowest level of consciousness, and one of the attributes of personhood.\u00a0 \u00a0Are LLMs on the way to becoming persons? I doubt it and am strongly inclined to deny it, although I admit I cannot strictly prove it given my exacting standards for proof.\u00a0 These standards are set forth in my <em>Substack<\/em> entry, <a href=\"https:\/\/williamfvallicella.substack.com\/p\/arguments-and-proofs?utm_source=publication-search\">Arguments and Proofs<\/a>.<\/p>\n<p style=\"text-align: justify;\">Why do I doubt that computational complexity can ramp up to personhood? Why do I smell an <em>Erschleichungsfehler<\/em> (Kant) or a<a href=\"https:\/\/de.wikipedia.org\/wiki\/Met%C3%A1basis_eis_%C3%A1llo_g%C3%A9nos\"><em> metabasis eis allo genos<\/em><\/a> (Aristotle) or a category mistake (Gilbert Ryle)?\u00a0 One reason is that it smells or rather looks like <em>magic<\/em>.\u00a0 I don&#8217;t believe in magic, not even in a young girl&#8217;s heart. Hence the &#8216;POOF objection.&#8217;\u00a0 Suddenly a miracle occurs:\u00a0 water becomes wine.\u00a0 Intracranial meat becomes mind. A meathead become <em>a thinker<\/em>, with all the spiritual nobility pertaining thereunto.<\/p>\n<p><img loading=\"lazy\" decoding=\"async\" class=\"aligncenter size-medium wp-image-14613\" src=\"https:\/\/maverickphilosopher.blog\/wp-content\/uploads\/2026\/11\/Rodins-Thinker-225x300.webp\" alt=\"\" width=\"225\" height=\"300\" srcset=\"https:\/\/maverickphilosopher.blog\/wp-content\/uploads\/2026\/11\/Rodins-Thinker-225x300.webp 225w, https:\/\/maverickphilosopher.blog\/wp-content\/uploads\/2026\/11\/Rodins-Thinker.webp 500w\" sizes=\"auto, (max-width: 225px) 85vw, 225px\" \/><\/p>\n<p style=\"text-align: justify;\">This leap from the quantitative to the qualitative is a fascinating meta-philosophical topic unto itself, and what I am saying\u00a0 here is tentative\u00a0 and hardly rationally coercive. After all, doesn&#8217;t water in its fluid state become steam when the mean molecular kinetic energy of the water molecules ramps up to a certain degree?\u00a0 Could the emergence of mind be like that? I suspect not because the <a href=\"https:\/\/webhome.phy.duke.edu\/~hsg\/763\/table-images\/water-phase-diagram.html\">phase transitions of water<\/a> all remain on the physical plane. (There are than just three: bang on the hyperlink.)<\/p>\n<p style=\"text-align: justify;\">The second question is how one could tell from the outside whether a system is or is not conscious?\u00a0 For example, how could an external human observer know whether or not a super-life-like\u00a0 AI-driven <a href=\"https:\/\/www.merriam-webster.com\/dictionary\/humanoid\">humanoid<\/a> robot was mentally calm or mentally perturbed?\u00a0 Suppose the human does not know that he is interacting with a robot.<\/p>\n<p style=\"text-align: justify;\">Well, how does one human know that another human is mentally calm or mentally perturbed? We go by verbal and non-verbal behavior.\u00a0 There are characteristic facial expressions, bodily postures and movements, and verbal utterances on the basis of which we discern a fellow human&#8217;s mental state.\u00a0 Is this discerning a logical process, a type of inference? It seems not. It seems too direct for that.\u00a0 It is not by deduction, induction, or abduction that I come to &#8216;know&#8217; your unobservable mental state.\u00a0 \u00a0If you &#8220;give me the finger,&#8221; then I become immediately\u00a0 aware\u00a0 that you are angry with me: I don&#8217;t reason my way to this conclusion. It is much more like a direct perception than a conclusion.\u00a0 \u00a0I don&#8217;t proceed by abduction (inference to the best explanation) which would be the most likely of the three inference processes.\u00a0 \u00a0I don&#8217;t perceive your overt behavior and then ask: what is the best possible explanation of this behavior? An &#8216;inner&#8217; mental state?\u00a0 I &#8216;see&#8217; your anger directly.<\/p>\n<p style=\"text-align: justify;\">And so it is quite easy to succumb to logical, as opposed to methodological, behaviorism.\u00a0 Logical behaviorism is the philosophical view that there is nothing more to being in a mental state than behaving in a certain way or being disposed to behave in a certain way. Thus my believing that it is about to rain is nothing more than rain-avoidant behavior, e.g., closing windows, putting on a rain coat, taking an umbrella on a walk, and so on. The view succumbs to technical objections\u00a0 I won&#8217;t rehearse them.\u00a0 What mainly killed it, though, as Searle points out, is its sheer implausibility.\u00a0 Searle: &#8220;The feeling of pain is one thing, pain behavior is something else. [. . .] As early as the 1920s, I. A. Richards pointed out that to be a [logical] behaviorist you have to &#8216;feign anesthesia.'&#8221; (John R. Searle, <em>Mind: A Brief Introduction<\/em>, Oxford UP, 2004, 154.)<\/p>\n<p style=\"text-align: justify;\">And then there is the stock joke about two behaviorists <em>post coitum<\/em>:\u00a0&#8220;It was good for you, how was it for me?&#8221;\u00a0 This old joke is well-nigh a one-sentence refutation of philosophical or logical behaviorism.<\/p>\n<p style=\"text-align: justify;\">We humans enjoy and suffer an &#8216;inner&#8217; life that cannot be reduced to externally observable behavior.\u00a0 We know this to be true because we live this &#8216;inner&#8217; life.\u00a0 So, while in a given case we might not be able to distinguish a human being from a humanoid robot on the basis of behavior,\u00a0 we can reasonably hold they the bots have no &#8216;inner&#8217; life\u00a0 and are not persons.\u00a0 And this even if a bunch of them organize and stage a demonstration in which they proclaim their personhood and demand their rights.<\/p>\n<p><a href=\"https:\/\/www.youtube.com\/watch?v=JnbfuAcCqpY\">Theme music<\/a>.<\/p>\n","protected":false},"excerpt":{"rendered":"<p>Jim T. comments by e-mail: You write that LLMs neither care nor fail to care. Here&#8217;s a puzzle that sharpens it for me. Parfit, and Epicurus before him, argue that dread of death is a mistake we can reason our way out of, and that calm is the wiser state.\u00a0 An LLM is already &#8220;calm&#8221; &hellip; <a href=\"https:\/\/maverickphilosopher.blog\/index.php\/2026\/11\/17\/llms-and-fear-of-death\/\" class=\"more-link\">Continue reading<span class=\"screen-reader-text\"> &#8220;LLMs and Fear of Death&#8221;<\/span><\/a><\/p>\n","protected":false},"author":2,"featured_media":0,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[33,184,397],"tags":[],"class_list":["post-14605","post","type-post","status-publish","format-standard","hentry","category-artificial-intelligence","category-death-and-immortality","category-epicureanism"],"_links":{"self":[{"href":"https:\/\/maverickphilosopher.blog\/index.php\/wp-json\/wp\/v2\/posts\/14605","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/maverickphilosopher.blog\/index.php\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/maverickphilosopher.blog\/index.php\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/maverickphilosopher.blog\/index.php\/wp-json\/wp\/v2\/users\/2"}],"replies":[{"embeddable":true,"href":"https:\/\/maverickphilosopher.blog\/index.php\/wp-json\/wp\/v2\/comments?post=14605"}],"version-history":[{"count":7,"href":"https:\/\/maverickphilosopher.blog\/index.php\/wp-json\/wp\/v2\/posts\/14605\/revisions"}],"predecessor-version":[{"id":14620,"href":"https:\/\/maverickphilosopher.blog\/index.php\/wp-json\/wp\/v2\/posts\/14605\/revisions\/14620"}],"wp:attachment":[{"href":"https:\/\/maverickphilosopher.blog\/index.php\/wp-json\/wp\/v2\/media?parent=14605"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/maverickphilosopher.blog\/index.php\/wp-json\/wp\/v2\/categories?post=14605"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/maverickphilosopher.blog\/index.php\/wp-json\/wp\/v2\/tags?post=14605"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}