This line of thinking traces to Drefyus’ What computers can’t do, and specifically of his reading of Heidegger’s care structure in Being and Time. Dreyfus’ views gained popularity during the first big AI wave and successfully put a lid on a lot of the hype around AI. I would say Dreyfus critiques are partly responsible for the terminological shift towards “machine learning” over AI, and also for the shifted focus on robotics and embodied cognition throughout the 90s. https://en.wikipedia.org/wiki/Hubert_Dreyfus%27s_views_on_artificial_intelligence But Drefyus’ critiques don’t really have a purchase anymore, and I’m surprised to see Sterling dusting them off. It’s hard to say that a driverless car doesn’t “care” about the conditions on the road; literally all it’s sensors and equipment are tuned to careful and persistent monitoring of road conditions. It remains in a ready state of action, equipped to interpret and respond to the world as a fully engaged participant. It is hard to read such a machine as a lifeless formal symbol manipulator. Haraway said it best: our machines are disturbingly lively, and we ourselves frighteningly inert. I think +Bruce Sterling underappreciates just how well we do understand the persistent complexities of biological organization. Driverless cars might be clunky and unreliable, but they are also orders of magnitude less complex than even a simple organism. The difference is more quantitative than qualitative, and is by no means mysterious or poorly understood. In a biological system, functional integration happens simultaneously at multiple scales; in a vehicle it might happen at two or three at most. This low organizational resolution makes it easier to see the structural inefficiencies and design choices in technological system. But this isn’t a rule for all technology. Software in particular isn’t subject to such design constraints. This is why we see neural nets making huge advances […]

There are two kinds of robot movies. The first treats robots as a spectacle. Robots in spectacle movies justify their existence by being badass and doing badass things. Sometimes spectacle robots work for the good guys (Pacific Rim, Big Hero 6). Sometimes they function as classic movie monsters (Terminator , The Matrix sequels) putting robots in the same monster family as zombies and Frankenstein, sources with which they share many tropes. But usually, spectacle robots serve as both heroes and villains simultaneously (Terminator 2, Transformers, Robocop, Avengers 2). Presenting robots in both positive and negative roles allows spectacle movies remain neutral on their nature. Robots can be a threat but they can also be a savior, so there’s no motivation to inquire deeply into the nature of robots as such. In effect, spectacle movies take the presence of robots for granted, and so reinforce our default presumptions: that robots exist for human use and entertainment. Robot spectacle movies can be entertaining but they tend to be shallow, and plenty of them are just plain boring (Real Steel, the animated Robots). Apart from functional novelties that advance the plot or (more likely) set up a slapstick gag, robot spectacle movies don’t bother to reflect on the robot’s experience of the world or how they might reflect on our human condition. The Terminator even provides the audience with glimpses of his heads-up display without hinting at the homunculus paradoxes it implies. Because once that robot’s function as a ruthless killing machine is established, the only question left is how to deal with it– a challenge to be met by the film’s human protagonists in an otherwise thoroughly conventional narrative. In spectacle movies, the robot is merely the pretense for telling that human story, another technological obstacle for humanity to overcome. The second […]

I’d invite everyone to imagine the human world as it existed before the invention of money. Prior to money, people engaged in cooperative behaviors for a variety of non-financial reasons (family, love, adventure, etc). But populations eventually grew too big to support the network with such slow, noisy transactions. Early agricultural societies invented money to help everyone collectively keep better track of how all their valuables were distributed. At the time, it would have been perfectly sensible to wonder about the complications that money would bring. “What if you’re in a situation where you need help, but you don’t have enough money to get anyone to help you? Seems like a lot of people could get the short end of the stick.” Of course, this worry would have been exactly correct. There are massive problems with the distribution of wealth and resources that comes with money. These problems are persistent; we still don’t know how to deal with them, and they are worse than ever before. Nevertheless, money was the critical coordinating infrastructure that (more or less) set up the human population to flourish over the last 10k yrs or so. It was the tool that built the human population as it exists today. I don’t like money. The human population today is fat, dirty, wasteful, uncoordinated in distributing resources, and ineffective at exerting global, targeted control that does anything but kill people. These failures have piled up to the point that they legitimately pose widespread, calamitous dangers to large human populations and important cultural centers. Money is currently in no position to resolve the problems we face; if anything, it’s made them virtually intractable. We’re in an analogous situation to the early agriculturalists: we need a new tool. Attention is our new tool; the attention economy our new coordinating […]

In a recent article, Adam Elkus argues two points: 1) Drawing attention to an issue doesn’t necessarily solve it. 2) Drawing attention might make things worse. For these reasons, Elkus argues against what he calls “tragedy hipsterism”: the “endless castigation of the West for sins and imperfections” without offering anything constructive. He says, “Awareness-raising is only useful if it is somehow necessary for the instrumental process of achieving the desired aim. In many cases, it is not and is in fact an obstacle to that aim.” I think this is completely mistaken, both about the utility of castigation, but more generally about the role of attention in shaping the social dynamics. Consider, for instance, a crying baby. Crying doesn’t solve any problem on its own. If an infant is hungry, crying won’t make food magically appear. At best, crying gets an adult to acquire food for the baby– but not necessarily so. The adult could easily ignore the baby, or misinterpret the cry as triggered by something other than hunger. Typically, an adult will feed the baby whether or not it cries, which renders the crying itself completely superfluous. And crying can be dangerous! In the wild, crying newborns tend to attract predators looking for an easy meal. On a plane, crying newborns create social animosity that might threaten the safety of the newborn and their family in other ways. Crying doesn’t always help, and it often makes things worse. So on Elkus’ argument, crying is actually an obstacle to the infant’s well being. If babies only understood the futility of crying, perhaps they’d be more effective at realizing their goals! Of course, this argument is ridiculous. Crying isn’t meant to solve problems directly. In fact, crying is usually issued from a place of helplessness: the inability to realize one’s […]

Andreas Schou writes: +Daniel Estrada finds this unnecessarily reductive and essentialist, and argues for a quacks-like-a-duck definition: if does a task which humans do, and effectively orients itself toward a goal, then it’s “intelligence.” After sitting on the question for a while, I think I agree — for some purposes. If your purpose is to build a philosophical category, “intelligence,” which at some point will entitle nonhuman intelligences to be treated as independent agents and valid objects of moral concern, reductive examination of the precise properties of nonhuman intelligences will yield consistently negative results. Human intelligence is largely illegible and was not, at any point, “built.” A capabilities approach which operates at a higher level of abstraction will flag the properties of a possibly-legitimate moral subject long before a close-to-the-metal approach will. (I do not believe we are near that point, but that’s also beyond the scope of this post.) But if your purpose is to build artificial intelligences, the reductive details matter in terms of practical ontology, but not necessarily ethics: a capabilities ontology creates a giant, muddy categorical mess which disallows engineers from distinguishing trivial parlor tricks like Eugene Goostman from meaningful accomplishments. The underspecified capabilities approach, without particulars, simply hands the reins over to the part of the human brain which draws faces in the clouds. Which is a problem. Because we are apparently built to greedily anthropomorphize. Historically, humans have treated states, natural objects, tools, the weather, their own thoughts, and their own unconscious actions as legitimate “persons.” (Seldom all at the same time, but still.) If we assigned the trait “intelligence” to every category which we had historically anthropomorphized, that would leave us treating the United States, Icelandic elf-stones, Watson, Zeus, our internal models of other peoples’ actions, and Ouija boards as being “intelligent.” Which […]

Scott Alexander: Should AI be Open? Or are we worried that AI will be so powerful that someone armed with AI is stronger than the government? Think about this scenario for a moment. If the government notices someone getting, say, a quarter as powerful as it is, it’ll probably take action. So an AI user isn’t likely to overpower the government unless their AI can become powerful enough to defeat the US military too quickly for the government to notice or respond to. But if AIs can do that, we’re back in the intelligence explosion/fast takeoff world where OpenAI’s assumptions break down. If AIs can go from zero to more-powerful-than-the-US-military in a very short amount of time while still remaining well-behaved, then we actually do have to worry about Dr. Evil and we shouldn’t be giving him all our research. // I’ve been meaning to write a critical take on the OpenAI project. I’m glad Scott Alexander did this first, because it allows me to start by pointing out how completely terrible the public discussion on AI is at the moment. We’re thinking about AI as if they are Super Saiyan warriors with a “power level” of some explicit quantity, as if such a number would determine the future success of a system. This is, for lack of a better word, a completely bullshit adolescent fantasy. For instance, there’s no question that the US government vastly overpowers ISIS and other terrorist organizations in strength, numbers, and strategy. Those terrorist groups nevertheless represent a persistent threat to global stability despite the radical asymmetry of power– or rather, precisely because of the ways we’ve abused this asymmetry. “Power level” here does not determine the trouble and disruption a system can cause; comparatively “weak” actors can nevertheless leave dramatic marks on history. Or […]