Posts Tagged ‘Plato’
“[They] would think that the truth is nothing but the shadows cast by the artifacts.”*…
How do AI models “understand” and represent reality? Is the inside of a vision model at all like a language model? As Ben Brubaker reports, researchers argue that as the models grow more powerful, they may be converging toward a singular “Platonic” way to represent the world…
Read a story about dogs, and you may remember it the next time you see one bounding through a park. That’s only possible because you have a unified concept of “dog” that isn’t tied to words or images alone. Bulldog or border collie, barking or getting its belly rubbed, a dog can be many things while still remaining a dog.
Artificial intelligence systems aren’t always so lucky. These systems learn by ingesting vast troves of data in a process called training. Often, that data is all of the same type — text for language models, images for computer vision systems, and more exotic kinds of data for systems designed to predict the odor of molecules or the structure of proteins. So to what extent do language models and vision models have a shared understanding of dogs?
Researchers investigate such questions by peering inside AI systems and studying how they represent scenes and sentences. A growing body of research has found that different AI models can develop similar representations, even if they’re trained using different datasets or entirely different data types. What’s more, a few studies have suggested that those representations are growing more similar as models grow more capable. In a 2024 paper, four AI researchers at the Massachusetts Institute of Technology argued that these hints of convergence are no fluke. Their idea, dubbed the Platonic representation hypothesis, has inspired a lively debate among researchers and a slew of follow-up work.
The team’s hypothesis gets its name from a 2,400-year-old allegory by the Greek philosopher Plato. In it, prisoners trapped inside a cave perceive the world only through shadows cast by outside objects. Plato maintained that we’re all like those unfortunate prisoners. The objects we encounter in everyday life, in his view, are pale shadows of ideal “forms” that reside in some transcendent realm beyond the reach of the senses.
The Platonic representation hypothesis is less abstract. In this version of the metaphor, what’s outside the cave is the real world, and it casts machine-readable shadows in the form of streams of data. AI models are the prisoners. The MIT team’s claim is that very different models, exposed only to the data streams, are beginning to converge on a shared “Platonic representation” of the world behind the data.
“Why do the language model and the vision model align? Because they’re both shadows of the same world,” said Phillip Isola, the senior author of the paper.
Not everyone is convinced. One of the main points of contention involves which representations to focus on. You can’t inspect a language model’s internal representation of every conceivable sentence, or a vision model’s representation of every image. So how do you decide which ones are, well, representative? Where do you look for the representations, and how do you compare them across very different models? It’s unlikely that researchers will reach a consensus on the Platonic representation hypothesis anytime soon, but that doesn’t bother Isola.
“Half the community says this is obvious, and the other half says this is obviously wrong,” he said. “We were happy with that response.”…
Read on: “Distinct AI Models Seem To Converge On How They Encode Reality,” from @quantamagazine.bsky.social.
Bracket with: “AGI is here (and I feel fine),” from Robin Sloan and “We Need to Talk About How We Talk About ‘AI’,” from Emily Bender and Nanna Inie.
* from Socrates “Allegory of the Cave,” in Plato’s Republic (Book VII)
###
As we interrogate ideas and Ideas, we might recall that it was on this date that the fictional HAL 9000 computer became operational, according to Arthur C. Clarke’s 2001: A Space Odyssey., in which the artificially-intelligent computer states: “I am a HAL 9000 computer, Production Number 3. I became operational at the HAL Plant in Urbana, Illinois, on January 12, 1997.” (Kubrik’s 1968 movie adaptation put his birthdate in 1992.)
“Memory is a wonderfully useful tool, and without it judgement does its work with difficulty”*…
Alexander Chee shares a memory…
It is the year 2004 and I take a seat at the counter of the Koreatown Denny’s, just three blocks from my apartment, and for a little while, I watch as a blonde waitress with makeup the colors of a tropical fish smiles at me every time she walks by. Her path is constant: she arrives from one side, departs from the other, grabbing or leaving pots of coffee on the warmer. She leaves a cup with me at my request and, in this way, I become part of the ritual.
I am a little drunk from drinks and no food. The day has become a kind of strange dream, telescoping down to the menu in front of me. I am here in Los Angeles for what will turn out to be seven months but I don’t know this yet.
At the counter, on one side of me are two young men studying a text in Spanish, the books so thick I assume they are Bibles. They ignore their pancake stacks. On the other side, a grizzled man of middle age sits, eating a hot fudge sundae.
Let me ask you a question, asks the man with the sundae.
Sure, I reply.
Is there ever a reason, a moral reason, to take a man’s life.
He spoons through the last bit of the hot fudge, putting it in his mouth. His hair, gray wire like a shoe brush; his glasses fish-eye his eyes. Say he is a judge, he says, and he sent you to prison for three years, didn’t allow you to have a fair trial. You know he had it in for you.
I look away from him and see that what I thought were the Bibles of the men next to him are Plato’s Dialogues, translated into Spanish. A sign that I might be in a Greek tragedy.
You do the time, the man continues. You get out. Would you have a right to take his life. A moral right.
I am God’s monkey, I think to myself. Watch me dance.
No, I say. Your duty after you leave prison is to yourself. I say this while looking forward, as if we are both in a car and driving. A moment later, I glance sideways and see the man’s wild eyes settle for a moment.
The reason you’re angry is because he didn’t value your life. To go and try to take his, that destroys what might be left for you in life.
But what if it felt good to do it, he says.
I note the use of the past tense. A confession? The waitress walks by again. Pleasure isn’t the highest value in this life, I say. Pleasure is only pleasure. It has no good or bad to it. That wouldn’t be a moral reason, at least.
This questioner takes it in. Hmm, he says. Thanks, he says.
To destroy him is to take some or all of what you have left and destroy it, I say.
He nods. Thanks, he says.
He pays and leaves.
Did he believe me, I wonder. I will feel a little more alone after that night in some way I will never understand and always try to forget.
Beside me now are only the two students of Plato. I order the sampler, it comes fast—mozzarella sticks, chicken fingers, onion rings. I eat them all…
From the annals of existential encounters: “The Denny’s on Wilshire Boulevard,” from @alexanderchee.bsky.social, one of his regular “I Come Here Often” columns in the LARB Quarterly (@lareviewofbooks.bsky.social)
* Montaigne
###
As we wonder, we might note that today is, appropriately to the piece above, National Chicken Fingers Day.
“I was conscious that I knew practically nothing”*…
As Joshua Rothman reminds us, we have a lot to learn from studying our ignorance…
… The truth, of course, is that we’re ignorant about the future. Who will win the election in November? Will we lose our jobs because of A.I.? Will the planet boil or merely simmer? What will skyscrapers, or smartphones, or schools look like in thirty years? We’re not in the dark about these questions; we can make educated guesses or predictions. But there’s an odd way in which, the more informed our speculations become, the more they serve to highlight what we don’t know. “The knowledge we possess determines the degree of specificity of the ignorance we recognize,” the philosopher Daniel DeNicola writes, in his book “Understanding Ignorance.” The more you know, the more precisely you can say what you don’t.
DeNicola’s book is an entry in a subfield of philosophy called “agnotology”—the study of ignorance. As philosophical subfields go, agnotology sounds abstract and even a little contradictory: what could it even mean to study what’s unknown? And yet, because ignorance is actually an everyday condition from which we all suffer, the study of it is quite down to earth. Have you ever been in a bookstore, leafed through a weighty tome, and then returned it to the shelf? You are practicing “rational ignorance,” DeNicola writes, by making “the more-or-less conscious decision that something is not worth knowing—at least for me, at least not now.” (In an information-rich society, he notes, knowing when to maintain this kind of ignorance is actually an important skill.) Have you ever tuned out a gossipy friend because you don’t want to know who said what about whom? Deciding that you’d rather be above the fray is “strategic ignorance”; you embrace it because it will make life better, deploying it when you decide not to read the reviews before seeing a movie, or conduct a hiring process in which the names of the candidates are obscured. There’s a big difference between strategic ignorance and what DeNicola calls “involuntary” ignorance: “In the iconic image, Justice is blindfolded, not blind,” he writes.
My wife’s parents have a box of letters that were sent between her grandfather and her grandmother while he was serving in the Navy during the Second World War. The box is in the basement; no one has read the letters, and no one plans to. This reflects a valid concern for privacy, but it also involves what DeNicola calls “willful ignorance”—the persistent, long-term maintenance of a gap in one’s knowledge that could easily be filled in. Willful ignorance isn’t necessarily bad; it might be wise to avoid learning the disturbing details of a half-forgotten traumatic event, for instance, lest they keep the trauma fresh. But we should be wary of willful ignorance, DeNicola argues, because it often flows from fear. “Consider a mother who is so upset about her son’s military service that she refuses to discuss it while he remains on active duty,” DeNicola writes. Or a voter who refuses to read about a favored candidate’s ongoing scandal. “The benefits of willful ignorance tend to be overestimated by those who exhibit it”; knowledge can be a path to overcoming fear.
DeNicola argues that, even when we don’t choose ignorance, there are ways in which we must “dwell in ignorance,” no matter what we do. We’re ignorant of most of what happened in the past because, despite our efforts at historical reconstruction, “worlds disappear” in the flow of time. We’re ignorant about the future not just because we don’t know what will happen but because we lack the ideas needed to comprehend future knowledge: “Galileo could not have known that solar flares produce bursts of radiation,” for example, because the very idea of radiation depends on a “framework of theoretical concepts” that wasn’t developed until hundreds of years after he lived. It turns out that there’s a special word, “ignoration,” which describes the condition of people who “do not even know that they do not know.” In a broad, almost existential sense, we all live in ignoration all the time. Recognizing this makes knowing what you don’t know feel like a step forward—even an opportunity to be seized…
… In a recent book called “Sense, Nonsense, and Subjectivity,” a German philosopher named Markus Gabriel argues that our personhood is partly based on ignorance—that “to be someone, to be a subject, is to be wrong about something.” It’s intuitive to hold the opposite view—to say that we are the sum of what we know. But Gabriel points out that, even when you know something to be true, you probably also know that there are aspects of it about which you’re probably wrong. I encountered this phenomenon recently when my son asked me to explain the meaning of “E=mc2”—but, also, when I tried to tell him about how I’d met his mom. “We were riding up in an elevator, and we started talking, and then she got off,” I said. “And then, later, when I was riding down, she got back on.”
This story is true, but also wreathed in inevitable uncertainties. What exactly did we say to one another? What were we wearing, or thinking, or feeling, before and after? There are limits to recollection, and to noticing in the moment; life is short, and you can’t know it all, not even about yourself. But you can know, at least to some extent, what you chose not to know and what you wished you’d found out. You can understand what you looked away from, and toward…
“What Don’t We Know?” from @joshuarothman in @NewYorker.
* Socrates, from Plato, Apology 22d
###
As we noodle on nescience, we might send bodacious birthday greetings to that most fabulous of flappers, Betty Boop; she made her first appearance on this date in 1930. The creation of animator Max Fleischer, she debuted in “Dizzy Dishes” (in which, still unevolved as a character, she is drawn as an anthropomorphic female dog).
“Men have become the tools of their tools”*…
Visionary philosopher Bernard Stiegler argued that it’s not our technology that makes humans special; rather, it’s our relationship with that technology. Bryan Norton explains…
It has become almost impossible to separate the effects of digital technologies from our everyday experiences. Reality is parsed through glowing screens, unending data feeds, biometric feedback loops, digital protheses and expanding networks that link our virtual selves to satellite arrays in geostationary orbit. Wristwatches interpret our physical condition by counting steps and heartbeats. Phones track how we spend our time online, map the geographic location of the places we visit and record our histories in digital archives. Social media platforms forge alliances and create new political possibilities. And vast wireless networks – connecting satellites, drones and ‘smart’ weapons – determine how the wars of our era are being waged. Our experiences of the world are soaked with digital technologies.
But for the French philosopher Bernard Stiegler, one of the earliest and foremost theorists of our digital age, understanding the world requires us to move beyond the standard view of technology. Stiegler believed that technology is not just about the effects of digital tools and the ways that they impact our lives. It is not just about how devices are created and wielded by powerful organisations, nation-states or individuals. Our relationship with technology is about something deeper and more fundamental. It is about technics.
According to Stiegler, technics – the making and use of technology, in the broadest sense – is what makes us human. Our unique way of existing in the world, as distinct from other species, is defined by the experiences and knowledge our tools make possible, whether that is a state-of-the-art brain-computer interface such as Neuralink, or a prehistoric flint axe used to clear a forest. But don’t be mistaken: ‘technics’ is not simply another word for ‘technology’. As Martin Heidegger wrote in his essay ‘The Question Concerning Technology’ (1954), which used the German term Technik instead of Technologie in the original title: the ‘essence of technology is by no means anything technological.’ This aligns with the history of the word: the etymology of ‘technics’ leads us back to something like the ancient Greek term for art – technē. The essence of technology, then, is not found in a device, such as the one you are using to read this essay. It is an open-ended creative process, a relationship with our tools and the world.
This is Stiegler’s legacy. Throughout his life, he took this idea of technics, first explored while he was imprisoned for armed robbery, further than anyone else. But his ideas have often been overlooked and misunderstood, even before he died in 2020. Today, they are more necessary than ever. How else can we learn to disentangle the effects of digital technologies from our everyday experiences? How else can we begin to grasp the history of our strange reality?…
[Norton unspools Stiegler’s remarkable life and the development of his thought…]
… Technology, for better or worse, affects every aspect of our lives. Our very sense of who we are is shaped and reshaped by the tools we have at our disposal. The problem, for Stiegler, is that when we pay too much attention to our tools, rather than how they are developed and deployed, we fail to understand our reality. We become trapped, merely describing the technological world on its own terms and making it even harder to untangle the effects of digital technologies and our everyday experiences. By encouraging us to pay closer attention to this world-making capacity, with its potential to harm and heal, Stiegler is showing us what else is possible. There are other ways of living, of being, of evolving. It is technics, not technology, that will give the future its new face…
Eminently worth reading in full: “Our tools shape our selves,” from @br_norton in @aeonmag.
Compare and contrast: Kevin Kelly‘s What Technology Wants
* Henry David Thoreau
###
As we own up, we might send phenomenological birthday greetings to Immanuel Kant; he was born on this date in 1724. One of the central figures of modern philosophy, Kant is remembered primarily for his efforts to unite reason with experience (e.g., Critique of Pure Reason [Kritik der reinen Vernunft], 1781), and for his work on ethics (e.g., Metaphysics of Morals [Die Metaphysik der Sitten], 1797) and aesthetics (e.g., Critique of Judgment [Kritik der Urteilskraft], 1790).
But Kant made important contributions to mathematics and astronomy. For example: his argument that mathematical truths are a form of synthetic a priori knowledge was cited by Einstein as an important early influence on his work. And his description of the Milky Way as a lens-shaped collection of stars that represented only one of many “island universes,” was later shown to be accurate by Herschel.
Act so as to treat humanity, whether in your own person or in that of another, at all times also as an end, and not only as a means.
“Will he not fancy that the shadows which he formerly saw are truer than the objects which are now shown to him?”*…
Bad news for those of us who occupy Plato’s Cave…
To all inhabitants of Plato’s Cave,
If you are receiving this letter, it means you have been designated a tenant of the cave—i.e., you are chained to the wall, you are forced to watch shadows for all eternity, you are projecting said shadow puppets, and/or you are a philosopher who was able to break free and understand the true shackles of reality (PhD candidates about to argue their thesis).
We are writing this memo to introduce ourselves, the new property managers of Plato’s Cave, and to let you know that this fall your rent will be raised.
We cannot wait to work with you as we journey together, but we also want to clarify that we mean real, actual money and not allegorical discussions on the concept of fiat currency.
- This is a memo and not a dialogue. You do not need to deconstruct the concept of reality or your interpretation of such. Please pay.
- Please be advised that we have changed our policies and will no longer accept imagined dialogues with Socrates and/or whimsical stories about Socrates.
- The rent will be used to address crumbling infrastructure as the upkeep of a completely underground cave is no easy thing. The money will go toward better walls, superior modes of imprisonment, a bigger and cooler fire, etc.
- To address any extra concerns, we will hold a fireside chat where you will be allowed to ask any remaining questions you might have. We understand you may not understand the “form” or “idea” of time, so we have allotted two weeks for this.
- Sure, you could break the chains of your supposed enslavement to the cave and embrace the complexity of reality, but consider we will offer Ice Cream Fridays. They will take place on the fourth Friday of every other month.
We do not undertake this lightly. As the costs of maintaining a cave meant to trap you in your ignorance increases year after year, we want you to know, from the bottom of our hearts, that we, too, are suffering. We get that times are tough, and we hope you can extend that sympathy to us, the managers of your cave…
Painfully funny: “Plato’s Cave Regrets to Inform You It Will Be Raising Its Rent,” from @Hellotherexu in @mcsweeneys.
* Plato, Republic
###
As we tighten our belts, we might recall that it was on this date in 1945 that farmer Lloyd Olsen of Fruita, Colorado, planning to eat supper with his mother-in-law and sent out to the yard by his wife to bring back a chicken, tried to behead a five-and-a-half-month-old Wyandotte chicken named Mike. The axe removed the bulk of the head, but missed the jugular vein, leaving one ear and most of the brain stem intact. The chicken was still able to balance on a perch and walk clumsily. He attempted to preen, peck for food, and crow, though with limited success; his “crowing” consisted of a gurgling sound made in his throat. When Mike did not die, Olsen decided to care for the bird.
Mike achieved national fame until his death in March 1947. In Fruita, Colorado, an annual “Mike the Headless Chicken Day” is held in May.
You must be logged in to post a comment.