These days, it seems like whether you are a big tech player in Silicon Valley like Google, IBM or Facebook or a small player in the middle of an irrelevant country like GoodAI (Prague, CZ - still EU), you are trying to develop artificial general intelligence (AGI). A year ago, I had a pleasant conversation with guys from GoodAI about the abilities their AGI should possess, and it still keeps me thinking. It seems like AGI is a topic for IT engineers, however, the good old philosopher may have a lot to say as well (yeah, I know, bullocks). The question 'what intelligence is' and how should we define it is old as Alan Turing's first ideas, but that is way too extensive theme for my brain today as it beautiful weather outside. I want to ask a simpler question: 'Is AGI going to redefine intelligence as we see it in humans in our everyday life?'
AGI developers and theoreticians focus mainly on the part of adaptation to the environment, creativity, and learning when they talk about the abilities AGI should have to be considered intelligent like humans. Learning: There is not doubt that computers are smarter than humans in some areas. The pocket calculator is way better in calculations than my brain is and that is a part of learning, at least as seen in our education system. You learn that 2+2 is 4. And computers can learn this too. So they can learn in some way. Creativity: I heard Margaret Boden talking about machines and creativity back in 2013 on 'HowTheLighGetsIn - The Philosophy and Music Festival in Hay,' where she stressed that machines could be capable of improbabilist creativity, which is a new combination of a known ideas, but cannot be capable of impossibilist creativity, which is done by broadening the known conceptual space, in simplification - they are completely novel. If that claim about impossibilist creativity is true, I would say, we will see probably in very near future. However, there is already some creativity seen in computers these days, at least the improbabilist one. Adaptation to the environment: Well, we all know these robot hoovers that can clean your room and they are capable of adaptations to the environment, don't we? These three capabilities are already praised in humans too. Creativity is given space in art galleries and museums; learning has been institutionalized through schools and universities, and we seek adaptation in children since they are born. But back to my question - what we can learn from computers that we are really good at because they will never be? We are born at one point in time, and we are going to die at another point in time, on the contrary, machines are not born, and they do not die. Hence, we have a limited time that is causally restricted by that death point. What I mean by that? I mean the basic things we do, is to avoid the death point. Obscure as it is, our lives are defined by death. So we have two time points, and these give us the notion of time, the notion of being and the notion of not being. That is something inherently human, and that is something why we ask: 'What is being? What is time? What is the meaning of life?'. So the only human capability that computers cannot learn is to be a philosopher in the sense we all are from our essence of being. And that is the ability we should praise as something exceptional. Yeah, I know, too deep for a short blog post. I wanted to argue that also our social behavior and knowledge are only human, but that would not be so remarkable. But they are. Can computer know, what is means to have a father? A wife? Why is my boss so frustrating? And one more thing not to worry about - we, humans, will be alone in our existential misery, Kierkegaard would say despair. Boden, Margaret A. (1994). Précis of The Creative Mind: Myths and Mechanisms. Behavioural and Brain Sciences. 17(3): 519-570. Turing, A.M. (1950). Computing machinery and intelligence. Mind. 59: 433-460.
0 Comments
I will start with quite a controversial topic. It might be risky, but it is better to be damned than bore readers to death right in the beginning. Yes, in the best case scenario both happen simultaneously. Social Situations Every time we appear in a social situation, our brain starts to engage millions of computational operations. For example, one of the cognitive abilities we incorporate is inference making about mental states of the others. This amazing cognitive capacity tells us, not only what the other think (with some degree of belief), but also what they feel, and how they are going to behave. This computation is based on our previous experiences and new evidence, that continuously changes our degree of belief about the other mental states. Two notes here: The experience does not necessarily mean our own lived experience, but also the unique aesthetic experience in the form of movies, books, theater plays, etc. Stories we heard from others count too, and even stories we make up ourselves. Here I talk about the degree of belief; I assume that this inference making can be described by Bays' theorem and that our social brain computations are Bayesian interpretations. The central brain regions involved in this "mentalizing" process are the medial prefrontal cortex (MPC), the temporoparietal junction (TPJ), and the medial parietal cortex. I have now summed some brain regions, but to be precise, the brain probably works as a network (as I and some other more qualified experts think, for example, Olaf Sporns). Therefore, in its entirety social regions most likely form a network that helps us overcome the awkwardness of many social engagements. When we approach someone, and we start a conversation, we need as many initial information as possible. To estimate the most probable inferences about the mental states of our companion, we have to determine his/her gender, sex, age, origin (through accents), look for other clues like clothes, gestures, tattoos, piercings, disability, ethnicity, make-up, hairstyle, and so forth. Yes, I said it - gender is relevant here, you do not mean bad, you just compute unconsciously. Gender is one quality that our brain process in social situations. These initial cues help to reduce the computational overload we encounter during social situations. You can imagine it as many equations our brain has to solve in seconds, and it is easier to solve them with as many numbers as possible, replacing the Xs and Ys with numbers. Now you see, knowing gender helps, but you might add: 'Wait, define gender.' (especially if you ever came across philosophy). Gender Stereotypes Here I refer to gender as a social construct that is being played, presented, implied, displayed, or done publicly. I chose the broadest form that can be represented by known gender theories by Bob Conell, Candace West and Don H. Zimmerman, Susanne J. Kessler a Wendy McKenna, Judith Butler, and Iris Marion Young. (Here, I do not want to look smart that I know all these names, I just want to show how many people are known for saying nearly the same.) Gender stereotypes are ascribed characteristics that refer to one being 'feminine' for women or 'masculine' for men. All possible human characteristics (and not only them, for example colors too) can be divided into the two complementary boxes 'feminine/masculine'. You learned them through gender socialization since your kinder garden like girls wear pink, boys wear blue, girls are nurses, boys are doctors. This categorization sounds trivial, but you most likely act according to those stereotypes without realizing it because you have your gender identity. Back to our social situation - why is it important to identify gender for mentalizing? Because almost certainly you are going to act/play/perform/do you gender role, and that is when I can fill in a number for X in my social equation and solve it. Gender is like playing according to rules. The rules can change, no doubt, but without them, there would be chaos, at least in your brain during social gatherings. In this post, I touched many interesting topics. Hopefully, I will come back to them in the future. Till then - love your Mr. Brain! Mitchell JP (2009) Inferences about mental states. Philos Trans R Soc Lond B Biol Sci 364:1309–1316. Renzetti CM, Curran DJ (1999) Women, Men, and Society. Boston: Allyn and Bacon. Sporns O (2013) Structure and function of complex brain networks. Dialogues Clin Neurosci 15:247–262. |
AuthorKaterina C. is a brain enthusiast, who also enjoys yoga (sometimes not, but it sounds cool), good beer (and the associated talks), and philosophy (which covers nearly every possible field). ArchivesCategories |
Proudly powered by Weebly