Huenemanniac

Here is a sensible, informed, and wise essay by Ali Minai at 3QuarksDaily that offers a clear picture of the general way AI systems like Chat-GPT work, and some sober recommendations for minimizing the disruptions they are likely to cause. His general conclusion:

The problem has to be addressed at least at three level: Regulation, engineering, and education. Regulation would control permissions, standards, rights, responsibilities, and accountability; engineering must focus on building systems that inherently pose smaller risks due to their design; and education must make human society more resilient to the potential harms of AI. None of these is likely to work perfectly, but their combination is the only way forward.

I recently had the chance to visit two centers of the arcane.

The first was the Warburg Library, an academic library at the University of London. It attracts a notable but somewhat rare subspecies of academics: those interested in how ideas in religion, philosophy, magic, mysticism, and the arts evolved from ancient times into the modern period. You might think this area of interest isn’t rare at all, and that a lot of historians get into this sort of thing. But my experience suggests otherwise. My sense is that most historians today are interested either in (a) questions that can be answered on the basis of data culled from archives (like figuring out where the timber came from to build the English ships that defeated the Spanish armada), or (b) superimposing what’s hot in our culture onto periods of the past (like exploring transgender politics in early modern London). It’s relatively easy to come up with something new to say in the (a) and (b) sets, so they can be taken up by capable PhD students and spun into careers leading to tenure. It makes sense that they would be more prevalent.

By contrast, the sort of learning that the Warburg encourages is old magic, special wizardy stuff. It requires putting together old texts from different periods, often in a handful of languages, in order to provide a comprehensive view of some strange or difficult idea that has animated people over centuries. These topics have been discussed forever, and more has been said about them than can be said intelligibly, so any scholar entering these thickets had better have hands full of time and patience, and a mind untroubled by rigid borders. It’s the style of thinking I admire the most, though at the same time I think it is wholly implausible. Basically, I don’t think there are any truths living at such high altitudes of time and space, and if there were, humans wouldn’t be in any position to know them. Still, it’s thrilling to watch some humans try, as they demonstrate the remarkable plasticity of the mind, if nothing else.

I’m fascinated by the Warburg as an institution because of its stubborn resolution to swim upstream against academic currents, but even more so because it has its own way of organizing its collection. The Warburg doesn’t arrange itself in some easy, straightforward way that allows a scholar to quickly lay their hands on whatever volume they think they’re looking for. Instead, it groups texts by a looser sense of association, as if the books all went to a pub and came home with interesting strangers. The library started with the personal collection of Aby Warburg, who read everything and shelved it according to his own sense of what relates to what, and the library tries to follow the same spirit today, so that scholars, looking for one book, will instead find other books they should be reading. I’m exaggerating only a little when I say the library is geared toward the production of new rabbit holes.

The other center of the arcane I visited was The Magic Circle in London. It is usually open only to members, but a friend of a friend gave us a tour. This is a place where great magicians gather to share secrets and discover new ones (so, rather like the Warburg). I am no magician myself, but I am fascinated by the intelligent techniques of deception and the knowledge of human psychology that is involved in stage magic. It is an art that requires long practice and deep dedication, but magicians (unlike many academics) rarely deceive themselves into thinking that their great efforts are carrying them into some sort of cloud cuckoo land. They keep their feet on the ground, and the cards up their sleeves. 

I feel that in my last two posts (on Tár and postmodernism) I managed to miss the big idea that was lurking in both of them. The idea is that the study of culture, and especially popular culture, is no substitute for understanding (what we might call) “the human condition” (if we didn’t know any better).

My brief and superficial post about approximating a definition of postmodernism was prompted by a book I am reading (Everything, All the Time, Everywhere, by Stuart Jeffries) that tries to explain the origins of postmodernism with a combination of economic policies and pop culture. The basic story, according to this book, is that, in the 1970s, neoliberal economics obliterated the modern dream of building a society that provided its members with fundamental goods (education, healthcare, protection against poverty, etc); at the same time, neoliberal politicians mouthed the ideals of freedom and the good life, with the rationale that untrammeled greed would provide everyone with a better lifestyle. Artists—and, especially, rock bands—responded by pointing out the manifest hypocrisy of both the modern promise and the neoliberal appropriation of that promise. They celebrated “western progress” ironically, pointing out the blatant lies and the cheap consumerism that was supposed to constitute human flourishing. This, basically, is what postmodernism amounts to: recognizing modernity for the fat lie that it is, and weaponizing its own false promises against itself.

There is much in this story I agree with, particularly the disasters that neoliberalism has generated and continues to generate. “Trickle down” economics really is just pissing on the poor, and all that. And rock bands and other artists have been pointing this out—at least until recently, as most of them by now have been co-opted by forces of capitalism. But putting these two things together (neoliberalism and rock bands) doesn’t really amount to anything like an interesting philosophical movement. I tend to agree with the economist Thomas Picketty that the neoliberal world economic order isn’t anything new: it is basically a return to how things were prior to WWI. And so protesting how bad this order is also isn’t anything new (see Marx, K.). I’m not sure there’s anything the Sex Pistols were trying to say that Rousseau didn’t say, and there’s something very 19th century about David Bowie. 

The deeper idea, I suppose, is that there is a fundamental conflict between some of the things we value in human life and our aspirations to create a system that provides those things or near substitutes. Humans have basic needs, and they like having loving communities of people looking out for each other; and they love creative expression, and they love some degree of conflict. They also hate being stifled by “loving” communities, and they hate being fenced in by rules that allow for greater population density (like “No loud music after 10 p.m.”). So they try to create systems that strike a livable compromise between meeting our human requirements efficiently but also allowing for individual expression. But every compromise proves unstable, especially when at the same time there are humans running around trying to enslave the others.

“Postmodernism is an intellectual stance or mode of discourse characterized by skepticism toward the “grand narratives” of modernism, rejection of epistemic certainty or the stability of meaning, and sensitivity to the role of ideology in maintaining political power.” [Wikipedia]

Postmodernism =

… a bunch of over-educated privileged urban types who like to make contradictory assertions and then defend them by insisting that nothing really means anything;

… a bunch of over-educated privileged urban types who like to make contradictory assertions and then defend them by saying that the meanings of claims always depend on who is speaking and what power they have over others;

… a bunch of over-educated privileged urban types who like to make contradictory assertions because by doing so they force the listener to think harder about what we say and what we mean;

… a bunch of privileged urban types who have studied books and artworks that undermine traditional assumptions, and who like to make contradictory assertions in order to prompt others to reflect critically on traditions;

… an intellectual and artistic movement among people who have learned to see through the sham-structures established by oppressive traditions, and who, through words and images, are pointing out the contradictions in those structures.

… an active critique of all the ways in which the assumptions of the past have infected our thinking, combined with an encouragement to think new things.

(OK, OK, it’s legit; but still, a lot of it is just smart-assery.)

I just came across this excellent overview of the field of linguistics by Steven Pinker. Highly recommended.

When I put together the “Two and a half minutes” piece for 3QD (below), I experimented with DALL-E to compose some art to accompany it. I ended up going with something less literal, but here’s what the AI did with the prompt “A painting of a muddy landscape with humans climbing out of brown pods”:

Looks like AI can pass the “think up some depressing imagery” test.

I have taught “Epistemology” for many years, but it has always been for me a difficult course to plan. I want to cover traditional philosophical questions about skepticism, justification, induction, and belief in the external world. But then I also want to cover topics arising from the social conditions of knowledge: how cultural ideologies and prejudices color what we perceive and what we think we know, and “the crooked timber of humanity” and all that. And then I also want to explore human psychology and our natural inclinations toward fallacious thinking, as well as how conspiracy theories arise, and the fresh challenges the internet brings to epistemology.

So finally, inevitably, I wrote my own textbook, and since textbooks are usually outrageously overpriced, I wanted to make mine an open resource ( = free!). I was lucky enough to gain tremendous, enthusiastic support from my university’s Open Educational Resource staff. And so here it is, for anyone interested!

Link to the book

At least on the surface, there seems to be something incongruous in regarding some artifact (like a watch) as clearly implying some kind of intelligent, crafty mind, but then regarding that intelligent, crafty mind as not implying any sort of further creator, but coming about through natural causes. A watchmaker is if anything more impressive and more in need of some explanation than the watch. So by what principle do we say that the watch requires design and artifice, but the watchmaker does not?

It’s a tricky question, because it is difficult to sort out “order that requires design” from “order that does not require design”. If we point out that all of an entity’s parts work together, that its behavior is regular or uniform, and that it is clearly not a simple heap or aggregate of parts, it is not clear whether we are talking about a watch or a giraffe. Natural objects and artificial objects all exhibit order; but which kind of order implies design, and which does not?

To sharpen the question a bit further, imagine the following case. Suppose we land on Mars and we find remarkable entities composed of stone. These entities can move themselves over the landscape, and they have parts. We observe that they are able to seek out specific kinds of rocks and assemble them into copies of themselves. Sometimes they break or wear down, but enough of them are able to make copies of themselves before this happens that there seems to be a steady population of them on the planet.

Imagine further that, in response to this discovery, the scientists on earth form two camps. One camp proclaims, “We have found life on another planet! The surprise is that living beings can be made of rock.” The second camp proclaims, “These robotic artifacts are evidence for intelligent robot engineers, who either once lived on Mars, or at least were able to send some clever rock robots to Mars.” One might initially say these are just two equivalent descriptions of the entities, but that isn’t so, because the two theories are very different in what they imply about the past: one theory says the robots arose naturally, without any designers, and the other theory says there must have been designers. So they are saying different things.

How can we tell which theory is right? Suppose we examine the entities more closely, and learn that they are powered by some chemical process involving the rocks and the Martian atmosphere, and that each one has the parts it has because their “parent” made them that way. They don’t “grow” or change on their own, except through erosion and minor collisions. They don’t heal, and, being rocks, they don’t have DNA. They don’t communicate in any way we can see, and each one of them, when put into a similar situation, “robotically” will do a similar thing. So far, no clear reason to regard them decisively as living beings or decisively as artificial robots.

In this imagined case, there is nothing inherent in the rock entities that tells us whether they are natural or designed. But there should be some fact of the matter, because these rock things must have come from somewhere, after all. 

This leads us to another possible way of answering the question. Could these rock entities have evolved through natural selection? Answering this question requires that we try to construct an explanation for them that appeals to some sort of random generating process, some sort of environment that selects for some entities over others, and enough time for these factors to result in the rock entities in question. The explanation we construct will have to be consistent with Mars’ history, so far as we know it, and should not rely upon too many improbable, lucky accidents (each one counts against the explanation). 

If it turns out that we can construct such an explanation, then it is possible that these rock entities have come about naturally, and if we have no real evidence for Martian or other-worldly engineers, then that’s the more likely explanation. If it turns out there is no way to construct such an explanation, then we will judge the design hypothesis as more probable (leaving some room for the possibility that future theorists will know more than we do, and come up with an evolutionary explanation). So there is a way to answer the question, and it comes down to what kinds of explanations we are able or not able to construct.

This offers a surprising lesson about the role of Darwinian evolution by natural selection in the distinctions we make between what is natural and what is designed. Indeed, the story suggests that “natural” just means “can be explained through evolution”–and old Charlie Darwin may not have realized that, in his work, he was in fact providing a criterion for nature itself! It also helps us to understand why “the design argument” held such powerful sway over intellectuals before Darwin: there was no clear criterion for when we should regard an ordered thing as natural, and so there was really nothing to keep anyone from seeing any ordered thing as designed. 

One of the cleanest and most compelling arguments against physicalism in the philosophy of mind is the “Knowledge Argument”. (Here is a quick summary. The response I am going to offer doesn’t show up there, though it fits in as a variant of the “No Learning” objection. It’s also the reply Daniel Dennett gives in Consciousness Explained.) According to this argument, thorough knowledge of the physical facts of a human being will not reveal any of the subjective states of that human being–what they feel, think, and sense. But this means there are facts about a human being that cannot be known by knowing all of the relevant physical facts. Hence, physicalism is false.

The argument is typically presented in a story about Mary. Suppose Mary is a super-smart brain scientist who is unable to see colors. She learns everything there is to know about the brain, including what the brain does when it sees colors. So Mary has all the neurophysiological facts. But then suppose she is given the ability to see colors. She sees a rose and exclaims, “Oh! So that is what red looks like!” She didn’t know what red looks like, even given her knowledge of the brain. Hence there was something Mary didn’t know about human conscious experience. Hence there’s more to it than physical facts can tell.

It’s a cute and somewhat compelling example. But it’s misleading, and the misleading part is the part in the story when it is claimed that Mary knows everything there is to know about the brain. Set aside the problem that that is quite a lot. The problem, really, is that that is not enough.

Suppose Mary is a super-smart leg-scientist. She knows everything there is to know about legs, including what legs do when they walk. One day someone asks her for the best route for walking to Las Vegas. She doesn’t know, as she has spent all her time studying leg physiology. Hence, there must be more to to walking to Las Vegas than just walking.

Well, yes, we should say, of course there is. One should consult a map of some kind, and it would be helpful to recommend sturdy walking shoes and so on. Just studying walking won’t tell you which direction to take. Similarly for studying the brain. The brain evolves and learns in a natural environment. Elements in the natural environment evolve as well, and sometimes in response to organisms’ abilities to process color information. The red stuff in the world tends to be stuff that commands attention, as red stuff is usually poisonous, or pretending to be poisonous, or yummy, or blood, or meat, or something else you should pay attention to. Part of what red is has to do with what things in the environment get perceived as red, and why. Color perception would not have evolved at all if it had not been useful for processing information.

So we should change the Mary case so that Mary knows even more. She not only knows everything there is to know about the brain, but also everything there is to know about colored objects in the environment, and what role they have played in evolution. So now Mary knows that seeing red evolved so as to alert organisms to threats and opportunities in the environment, and that seeing red things usually results in a charged experience–scary, appetizing, sexy, etc. Red is attention-commanding. As such, it had better stand out brightly against things that can usually be ignored in crisis situations – green and brown things, for example. Mary may as well learn all the ways that red and other color words have been culturally embedded as well, in poems and stories and religious ideas, so as to understand the extensive role red plays in human experience.

Now once Mary knows all that, is it as obvious that she wouldn’t know what red looks like? I’ll admit, it may seem like she still wouldn’t know exactly what it looks like. But I don’t think it is obvious that she wouldn’t know. She might well gain her color vision, see a red thing, and say, “Ah, that’s pretty much what I expected!”