Philosophy and AI

I think AI will be good for philosophy (if the philosophers don’t ruin it first, a big if). The reason is obvious: AI is no good at philosophy, at least as AI now exists. But it is good at many other things: white-collar jobs, routine teaching, performing calculations, storing information, providing quick answers. It is good at math and science, but not good at philosophy. So, jobs in fields where AI is good will disappear, but philosophers won’t. This is all commonly accepted, more or less. But I think it goes further—into news and entertainment. I think animated newscasters are a distinct possibility, along with robot cameramen and computerized news writers. I’m not sure about comedians (see “The Comeback” on HBO), but I predict that actors are going to be in trouble, especially action stars and leading men and women. For the powers of animation are going to take over their roles (literally). We already see it happening in sci-fi and fantasy films, also pornography. Nature documentaries are on the brink of extinction. AI is just less expensive and more flexible. You won’t even need scriptwriters for the commercial stuff (are there any real writers in Hollywood anymore?). A dumbed-down culture like ours is an AI culture.

But philosophers are not expendable in this way, except for routine teaching. The only question is whether the demand will exist. If universities as we know them succumb to AI, partially or wholly, philosophers will go elsewhere and form companies—so long as there is a demand for them. This is the big question: in an age of AI will people become stupider or will they aspire to something more elevated—difficult, challenging, profound? My suspicion is that they will do the latter: their minds will be less taken up with routine mechanical tasks and freer to roam more widely and deeply. Philosophers will be in demand, treated as valuable commodities. Movie stars will fade away, and TV celebrity pundits, to be replaced by philosophy stars. AI will come to them. I use AI to do my grunt work, but I don’t use it to make progress on philosophical issues, and I never will (not in this lifetime). I envisage a society in the not-too-distant future in which philosophers (good ones anyway—there are plenty of hacks) will be regarded as superstars. Even the architects of AI will lose out to AI, as AI figures out how to manage and create AI. Even in a society dominated by AI, not always benignly, philosophers will still be necessary—and I am really not so sure about scientists, historians, and even athletes. Armies won’t need flesh-and-blood troops (we will have robotic boots-on-the-ground and drones in the air). In an AI world actual organic philosophers will be kings, or at least “Hollywood royalty”. And the underlying reason, in my book, is that we philosophers deal with insoluble problems, while AI deals in soluble problems (problems not mysteries). It will be good for us that our problems can’t be solved.

Entanglement Epistemology

Is there any alternative to empiricism and rationalism—experience and reason, impressions and innate ideas?[1] In particular, do we need to have experiences of everything we know? Can empirical knowledge (as opposed to the a priori kind) be constituted by anything other than sense impressions of the thing known? One might hope so, because many things in the world are not directly experienced. Can we just know a thing without needing to experience it? What else could provide knowledge apart from experience? I am going to suggest that the empiricist conception of empirical knowledge is fundamentally mistaken; such knowledge has a quite different nature or essence. The concept of experience is not the right concept to capture it. I will build up to this slowly.

We already have ideas floating around that imply a non-experiential conception of knowledge. One idea is correspondence or correlation: to know a fact is to have a mental state (possibly a belief) that corresponds to a fact or correlates with a fact; here is the belief and there is the fact, and the two are in a relation of correspondence or correlation. There is a mental bit and a worldly bit, and one bit fits the other: my belief that snow is white corresponds to, or is correlated with, the fact that snow is white—so I know that fact. But this simple theory runs into trouble because the two need to be connected not just correlated: it has to be becauseof the fact that snow is white that I believe that snow is white. Notice that this theory says nothing about experience; it is more abstract than that. The natural next step is to introduce the concept of causation: to know a fact is to be in a state that is caused by that fact (or a related one). Thus, we obtain a causal theory of knowledge: all empirical knowledge is caused by external facts; there is no non-causal knowledge of the world. Instead of saying that all knowledge depends on experience, we say that all knowledge depends on causation (then we go on to specify what kind of causation). This is a non-empiricist theory of knowledge (also non-rationalist). It may not be much good as a theory, but it isn’t empiricist in the classic sense; experience doesn’t come into it and may not even be present in the knowing subject. It can handle knowledge acquired unconsciously, as in subliminal perception and blindsight, and it doesn’t require that all knowledge be based on direct experience of the thing known. It allows for knowledge in beings that lack sensory experience (“zombies”) but are still hooked up to reality causally. It makes room for the possibility that conscious experience is epiphenomenal and so never the cause of empirical beliefs (brains states are). It puts experience in its place epistemically.

We can sophisticate this simple causal theory, keeping its anti-empiricist flavor. Thus, we get reliability theories: to know a fact is to be reliably connected to facts in general. This ensures that the truth of one’s belief is not just a fluke; the subject is reliably right about relevant things (remember the red barn example). He is reliably connected to the facts. All knowledge is based on reliable connections between beliefs and facts, possibly backed by causal connections. Knowledge is reliable (causal) connection. Nothing experiential has been mentioned here. In the same vein we might bring the concept of impingement into the picture: knowledge arises from, and depends upon, the impingement of the world on the mind.[2] This is the essence of knowledge, its conceptual heart. And it is suitably general and capacious: we can allow for unconscious knowledge and knowledge of things we can’t directly experience (as well as blindsight knowledge). We thus improve upon classic experience-based empiricism; this is a kind of non-experiential empiricism (denial of rationalism). All knowledge is acquired post-natally by means of worldly impingement, but experience (a certain kind of mental state) need play no role. If you don’t believe in experiences, you can still be a kind of empiricist—you can still believe that all knowledge comes from interactions with things outside the mind. You may admit that the theory is rather crude and oversimplified (and promise to work on improving it), and you may also accept that knowledge is somewhat mysterious as so conceived; but you are certainly not a classic experience-obsessed empiricist, like Locke and Hume. You reject the notion of phenomenal experiential foundations (the “given”) and inferences therefrom. Yet you hold that knowledge is all about being impinged upon by the world outside. Experience comes into it, if it comes in at all, only per accidens; it is just one way of being impinged upon. You are what might be called a connectionist about knowledge not a phenomenalist: knowledge is all about objectively connecting to external reality, not about subjectively experiencing from within. Inner feeling (sensation) is strictly irrelevant to knowledge per se. What your mind can connect to determines what you can know, not what can come before it phenomenally.

Where does entanglement come in? Perhaps it will surprise (and alarm) the reader if I say that the idea of quantum entanglement is being invoked. For I wish to suggest that knowledge is a form of entanglement in roughly the sense that particles can be entangled in quantum physics. That is, particles can be inextricably linked to other particles in ways that defy easy comprehension—yet they are so linked. Similarly, minds can be inextricably linked to other things in a curious and sometimes baffling manner—yet they are so linked. For example, your mind can be connected (entangled) with dispositions, possibilities, space, time, the self, other minds, and moral values, in ways that defy ordinary conceptions (mainly mechanical conceptions). The entanglement is mysterious relative to our intuitive commonsense ideas about how things work in nature (like gravity). But it happens; it’s real. The OED gives us “to make tangled” for “entangle” and for “tangle” we have “twist together into a confused mass”. But things can be twisted together otherwise than into a confused mass, e.g., braiding and rope. The word “entanglement” in physics does not connote confusion or chaos (it is perfectly lawlike), but it does signify a kind of spooky joining together or union. Just so, mind and world get yoked together in instances of knowledge, improbably and weirdly. Your thoughts get linked to dispositions (say), even though it is hard to see how you can properly conceive of their existence (because of their imperceptible counterfactual implications). The mind ought not to be able to form such thoughts (and hence knowledge), but evidently it does. You can think about space and time, even though you are baffled about what they are. You can know other minds, but by rights you ought not to be able to (you have no direct experience of other minds). Somehow you can know about numbers, but have never seen one. There is epistemic entanglement of a puzzling (indeed uncanny) nature. It’s not like balls of string, where you expect them to get entangled. There is a strange kind of connection that contravenes empiricist principles (“no knowledge without direct experience”). Knowledge is a kind of mystifying connection (compare intentionality).

This knowledge just happens, even if we don’t know how or why. In effect, we have mysterious epistemic connectivity. Classic empiricism purported to give us a clear, intelligible, almost mechanical, picture of how knowledge is formed and constituted; but in reality, it is more mysterious and mind-stretching than the empiricists thought. It is a puzzling entanglement, but real enough. As a consolation, knowledge is more easily obtained than empiricist principles allow: there is an awful lot we don’t know according to these principles, which we patently do know. It is fair to report that empiricism has been regarded as a bad theory inasmuch as it fails to account for a good deal of human knowledge—but it has been reluctantly accepted largely because no other theory suggests itself. But now we see that there is another theory that is more empirically adequate, though less pellucid. And how pellucid is classic empiricism anyway—what is experience, and how is belief based on it? How exactly does the alleged “inference” work? We have to rethink human (and animal) knowledge from the ground up, recognizing its peculiarities. Epistemology is as perplexing as quantum physics (well, maybe not quite as perplexing). We need an entanglement theory combined with a dose of mysterianism. Mind and brain are connected, despite the puzzling nature of their connection; mind and world are also connected in the phenomenon of knowledge, despite the puzzling nature of the connection. We should be connectionists about both, though perplexed connectionists.

But we shouldn’t feel totally defeated in the case of knowledge, since it is quite evident that we know various things that our traditional theories say we can’t know, e.g., that we have specific sensory faculties. Knowledge is eminently possible, despite skeptical protests. I just do know stuff that I can’t directly experience, okay. Not everything I know about gives rise to sense impressions in me (atoms, numbers, selves), that’s a fact. In cases of knowledge mind and world get spliced together in remarkable ways, get used to it. Epistemologically, we exist in an entangled state of nature. The thoroughly modern empiricist should be saying that all knowledge is a result of epistemic entanglement with nature—not derived from hallowed tradition or God or our omniscient genes. Experience might come into it at various points, but it is not the be-all and end-all. The basic concept of knowledge is not an experiential concept. My knowing mind can become epistemically entangled with color by perceiving it, but it can also become entangled with the faculty of sight by my possessing that faculty, by embodying it. Similarly, my mind can get tangled up (twisted together, intertwined) with dispositions, space, time, selves, numbers, values, etc., without perceiving them with any of my five senses. We are thus not condemned to an experiential reduction of all knowledge in the manner or classic empiricism (or its more or less faithful descendants). Knowledge is what it is and not some other thing.[3]

[1] This paper follows on from my “Faculty Knowledge”.

[2] See my “A New Definition of Knowledge”.

[3] In case you haven’t noticed, this paper is a rapid-fire survey of the history of epistemology up to the present day. The aim is to reinvent the subject by forging a new path. We are tremendously in the grip of traditional empiricism (less so rationalism) and it takes an effort of will to see through it. I intend my use of the word “entanglement” to create a new way to approach the subject, though a way not without precedents. We are always looking for the right words, and the English language will only take us so far. Knowledge is not easy to describe, or even conceptualize—like many things.

Faculty Knowledge

Empiricists and rationalists agree that we have a number of cognitive faculties: the five senses plus whatever else is needed to account for the totality of human knowledge (mathematics, ethics, etc.). But how do we know this—how do we know, for example, that the visual faculty exists? The empiricists and rationalists never discuss this question. But we can imagine their answer: by experience or innately. I know that I have a visual faculty (or a mathematical faculty) by experiencing it with my senses or by dint of my genetic endowment. I have an idea acquired by the senses or derived from my inborn nature—post-experience or pre-experience (a posteriori or a priori). I know it by perceiving it or by pure reasoning—by sense or reason. I know that the faculty exists this way and that it has certain characteristics—the knowledge arises by copying sensations or by employing the innate system. That is, I know my cognitive psychology according to one method or the other, according to epistemological predilection. But are these theories plausible? On the face it, they are not: I don’t have sensory impressions of my visual system (or my mathematical competence), and I don’t know it innately either. I don’t see my ability to see with my eyes and I don’t know it with my genes either. I don’t experience my cognitive faculties (as I experience the color red) and I don’t know them implicitly in the womb. I know them in some other way. But what way?

Danger lurks, epistemological danger. Either we have no theory at all or we have no knowledge of the kind in question: no explanation or no knowledge. Either we don’t know how we know our cognitive faculties, or we don’t have any knowledge of them at all. Empiricism and rationalism seem exhaustive, and skepticism is a last resort. Surely, I know that I have a faculty of vision, and yet we have no theory of how this is possible. You might think we know it by the old standby of copying: I have a visual experience and I copy this into an idea that counts as knowledge. That is, I know it by introspecting my experiences. But that can’t be right because an experience is not a faculty: I could introspect my visual experience but not know what vision is—that thing I do with my eyes. To know that I can see goes beyond being acquainted with my visual sensations; for seeing is an ability, a competence, a psychophysical capacity. The same is true for my other cognitive faculties, like my mathematical ability—I don’t just introspect my mathematical experiences; I know what I can do mathematically. When I know that I have a linguistic faculty, I don’t just know what words sound like. So, the empiricist theory, in its classic form, looks hopeless as a matter of principle. On the other hand, it sounds farfetched to claim that I have such knowledge innately: that I am born knowing the existence and nature of my various cognitive faculties. Somehow this knowledge makes its appearance later, as a result of…as a result of what exactly? I seem to just know it. I know that I have the ability to see (discounting the depredations of skepticism—the other option).

Here we might follow a traditional path: we weaken the empiricism to make it more tolerant of the limitations of direct perception. I infer the existence of my faculties by some method of theory construction, say inference to the best explanation (otherwise known as shooting in the dark); or I settle for having a theory that has not so far been refuted and that I like the sound of. I won’t go into the merits of these maneuvers; I will only remark that they don’t do justice to the primitive immediacy of the knowledge in question. I’m not guessing I have the ability to see or speak or calculate—I know it. It’s not like speculating about atoms or distant galaxies or dinosaurs. Thus, I know something that the standard theories can’t explain—I have faculty knowledge. Nor do we have any other theory to put in their place—do I know it magically, unaccountably? The knowledge looks suspiciously like telepathy—self-directed telepathy. I am just primitively and mysteriously aware of my cognitive faculties (mysterianism beckons). Maybe there could be knowledge that eludes our comprehension in this way, but what is troubling is that empiricism and rationalism seem like the only games in town—where else could knowledge come from? Nothing else could explain the origin of knowledge; we know these are the only two theoretical possibilities.

The irony is that we can’t explain the origin of knowledge of something that is supposed to explain all our knowledge. The sensory and rational faculties that define us as a knowing species have no intelligible explanation as to how they are known. This kind of knowledge is a counterexample to both empiricism and rationalism, or the conjunction of both. It is not knowledge by experience and it is not knowledge by innate endowment—yet it is really knowledge. Ouch. It might be thought that there is a way out of this conundrum, though hardly a cheering one, namely that it is a special case of knowledge of dispositions. That is, we know our cognitive faculties as we know dispositions in general, e.g., that salt is soluble in water. Knowledge of dispositions is notoriously difficult for empiricism, since possibilities (what things would do if) are not perceptible—dispositions don’t allow knowledge at all, just unjustified faith. It would follow that our faculty knowledge isn’t really knowledge at all—just conjecture. And yet that sounds completely wrong. Nor can rationalism save the day, since it is highly implausible to claim that we know that salt is water-soluble innately. Dispositional knowledge is difficult for both empiricism and rationalism. But a skeptical response will not serve the turn in the present case, since it undermines the theories it was intended to bolster—viz. that we know that empiricism or rationalism is true. For, if we can’t know that we have the faculties in question, we can’t know that all knowledge depends on them; we can only say that if these faculties exist, then they explain all knowledge. That is a dialectical disaster of epic proportions. It is self-refuting. We have to know that we have senses in order to assert that sensing is the origin of all knowledge, so we can’t admit that we don’t know that we have senses. The empiricist just assumes we know we have senses, especially vision, but this knowledge is not subject to the basic empiricist principle, viz. that all knowledge is based on experience. He was right not to bring this case up at the beginning, because it would have given the game away—he can’t even explain the existence of knowledge of the very faculties on which he is depending to explain all knowledge! He can’t, for example, explain how we know we can see, i.e., have a visual sense—in the ordinary sense of that phrase. Vision is something in the world that we can know, but empiricism can’t explain such knowledge (and neither can rationalism). So, we are in want of a general theory of human knowledge such as these two great epistemologies were intended to provide. The simple fact is that faculty knowledge violates both empiricism and rationalism. No wonder no one ever talked about it.[1]

[1] I do think we should consider the possibility that both empiricism and rationalism are both completely on the wrong track as generaltheories of knowledge, though the entire tradition is steeped in their precepts. There might be a way of knowing that simply doesn’t slot into these schemes, but employs quite different mechanisms. I am encouraged in this thought by the evident mysteries that surround human knowledge—mysteries of empirical knowledge (so-called) and mysteries of innate knowledge (also so-called). We really don’t understand how sense experience causes knowledge and we don’t know how innate genetic endowment can constitute knowledge (see my Problems in Philosophy and Inborn Knowledge). Our understanding of self-knowledge is particularly poor—is there any science of it? Isn’t knowledge that I can see a paradigm case of knowledge? And yet we don’t know how it’s possible. I don’t perceive my ability to see with my senses, or conjecturally construct this knowledge from what I perceive with my senses; and I don’t know it innately either. So, how do I know it? We have a kind of skeptical paradox here: what does this knowledge consist in?