A New Philosophy on Artificial Intelligence | Kristian Hammond | TEDxNorthwesternU

thank you, thank you all. I deeply appreciate you being here on a beautiful day to be in a dark room listening to an academic. I don't know what they were thinking. Today I'm going to talk about machines, in fact, in particular. I am going to talk about two machines, computers and people. The idea here is that I'm really going to talk about who we are from the perspective, in many ways, of what we are doing now with computing and I think one of the most important thing to start with is realizing who I am and that is that I build smart machines.

I'm a computer scientist, but my focus is

artificial

intelligence

, which is building machines that think like us in particular. It's important that you understand that. since I was a little kid I love the idea of ​​smart machines smart robots smart computers things that would help us things that would try to kill us and I love them even in those times when people said they were trying to kill us and I think at that time I was one of the few people who, after watching a Space Odyssey 2001, actually thought that after trying to destroy the entire ship he was on, what about all the people on that ship who got a bad reputation? ? he was evil, he wanted to achieve his mission and by the way no one told him while you are achieving your mission don't kill everyone but we have an interesting world ahead of us now and what I care about is watching.

In this world there is the issue of

intelligence

in all its forms because I really believe that there are three big questions in the world. The first is what we have seen for years and it is the note or the physical world. and chemistry, how things work at the atomic and molecular level, the second is the nature of life and how biology, evolution, ecological systems emerged and the third is intelligence, the nature of intelligence, the physical world , the biological world, the intellectual world. are the three big questions that we have in front of us and from my perspective, the problem is that cognition is cognition, it doesn't matter if it's human cognition, it doesn't matter if it's machine cognition, the only thing that matters is that it is cognition.

Now it turns out we have a horrible relationship with machine intelligence, I mean a really devastating relationship with machine intelligence, most of it has to do with bad press, but isn't it? We have a hard time thinking about intelligent machines, first of all, we think They're going to kill us, which is fine, I mean, we've been told this for years, from Colossus to Terminator, we've seen killing machines and, you know, in the movies of all our lives, more recently we also have this notion that they are going to take away our jobs, that is, they are going to take over everything we can do and we will be left alone in the dark with machines running everything.

One of the ways we fight, we doubt it. Do we think of ourselves as different and unique and that there will be absolutely nothing that the machine can do or that we get the machine to learn to be absolutely nothing that we can do to build machines that can do everything? of the things we can do and then we stand up for ourselves and look directly at examples of our own decision making and talk about the machine that makes those decisions in terms of being a coal, being cold and calculating and having absolutely no emotion or effect or any thing while we develop deep and fundamental emotional relationships with things because the reality is that we are terrified by the idea that we are able to program something that is like us because if we can program something that is like us then maybe they are programmed or programmable and therefore We fight all the time to protect what is different and unique about us, but what makes us special?

What do we have that the machine will never be able to do well? a lot of things that we started with and when we think about our battle with the machine and I think you've probably heard most of them, is the idea that, well, machines will never have emotion, or creativity, or humor, or intuition, or consciousness. but consciousness is always interesting because whatever you say, and this machine became conscious, the next sentence is and then it tried to kill us all, it's always like that, but the thing is, the other ones, the other top three, are interesting and important, but today I just want to talk a little bit about intuition and consciousness because those are the things that we really consider fundamental and that we all have, we are able to make decisions intuitively and we are aware of the world.

Now there is a kind of irony here. in terms of these two characteristics that are our protection in terms of the fact that we are human beings and that is that intuition is the ability to make decisions without knowing very well why we are making a decision in consciousness, consciousness is the notion that we always we are aware of our thoughts we are aware of our thoughts we are aware of the external world that we are in between us, but that means that we are in this curious position where we are not aware by definition of the processes that give us intuition so these two are absolutely opposite sides of the coin of the reasoning that shapes us so let's think about intuition for just a second we all know what it is to follow our instinct we don't have enough evidence that we are going to have to make an assumption and in fact it is often the case that there will be that tension between people who are logical and people who are intuitive who say: I don't want to listen to the numbers.

I go with my instincts. I don't mind. On the evidence, I'm going to do what I'm going to do. I've seen this thousands of times before trying it, and in fact, almost every genuinely famous and successful person has said something about intuition, about how wonderful intuition is how fantastic intuition is. There's something called survivor bias and that's when you look at people who are doing very well and you ask what characteristics are really associated with them doing very well and a lot of them will say intuition of Of course, intuition is also a characteristic associated with people who make bad guesses.

It turns out that smart people make very good guesses and not-so-smart people make bad guesses, but it all comes down to making a decision. an understanding of what is happening in the world, evaluating what is happening without knowing why and we look at the Machine, we think that what machines don't do, that they are all based on rules, but that is not the case, machines They are already incredibly intuitive, I mean, amazingly intuitive. That is, much of the work we are seeing today in the world of what is called deep learning has to do with developing machine intuition, that is, giving a machine example after example of an example and learning slowly from everyone. those examples until you get to the point where you can look at a new situation and say know what's going on here and find an answer and this is what we see when we have things like image recognition on Facebook, Siri can understand your words. the autonomous car knows how to adapt to the road.

Now the amazing thing is that these machines are intuitive in the sense that they can make decisions but they can't articulate why they make those decisions. There are no rules anymore. Thinking networks that we have built and that we train so that people have the machines and can make these decisions, but the question is: do we want machines that do that? Do we want machines to be able to make decisions without being able to explain themselves in general? The answer is no, if I have a self-driving car that runs over someone to save someone else's life, I would actually like it to be able to explain well if I have a system that makes a decision about someone's life. future I want to be able to explain myself so for us we look at machines that make decisions without knowing why this seems strange and the question is when we look at people maybe we should consider that that is also strange, that is, the ability to make a decision without Understanding why you are making a decision may be a human ability, but perhaps we should aspire to make the decisions we make our own, we make them understandable, the explainable, they are taken to a different type of realm that is not instinct but brought back to the brain along with intuition, the other side of this is consciousness, unconsciousness is very special to us, it is this notion that we are aware that you all have at this moment a conscious awareness of what you are seeing, what I'm saying. where you are sitting what you are thinking there are other things going on in your head besides what is going on in the outside world and this is what being means to you, i.e. we are looking at the outside world we are looking at the things inside of us and we are hooked to those things, so it is this awareness of the internal and the external.

Another thing is that, as we watch, the question is what does this really mean in our lives. Is this awareness what it does for us? What role does it play? And I say, as things that exist in the world, what does it mean to me to be aware of ourselves? So the key point to understanding this is push-ups, so if you do 50 push-ups. Every morning for the rest of your life you will be healthier and live longer. The way I think it's fair is that once you decide that you say, "I'm going to do 50 push-ups every morning for the rest of my life." The next morning you'll wake up and you'll be fine, you know, I have things to do, I do, I have a brush, but I have a shower, I have to go down, oh my God, I'm late, no.

I'm going to do it and the next morning it's like hey, I'm here, so you're in this weird situation where you know you want to do something and you don't want to do something yet, but if the part of you that wants to do it pushes - oops. gets a little aggressive and tells the part of you that says say no today we are going to do pushups damn you are doing it 50 pushups today now you do the feet part today dinner onion you forced that part of you to do 50 pushups a day next and the next day and the next day for 30 days or 45 days, then the part of you that didn't want to do push-ups wants to do push-ups now that it knows you're supposed to do them. get up in the morning and even when you don't want to do pushups anymore but it says today we're going to do pushups, damn it, but we're going to do pushups, let's say you can train yourself to do pushups every day, which is a little strange in terms of a speech, you're training like there are two parts of you and then you think about the memory, you've all had a moment where you couldn't remember someone's name, it's like who played who played who played Commissioner Gordon in the movies Christopher Nolan's Batman this I don't remember his name he played he did he just did Churchill won an Oscar I guess he did Sid Nancy years ago oh it's Gary Oldman I mean though you have a memory and sometimes when you do a ask your memory, it can't come up with the answer, but you know how to manipulate that question and ask it more and more questions until you get an answer, i.e. memory.

It's not exactly you, you're aware of memory, but it's not exactly you, it's something that you can use and I would say the same thing when you're thinking about driving a car because sometimes when you're driving a car, you kind of go. So you forget about it. that you're just driving, you're not thinking about what lanes you're moving in, you don't know anything about what's in front of you, but something is taking care of you, especially if you have a route to go. Every day you'll just drive it and you'll be able to think about other things while you're driving and we have these activities that we do that we don't think about, we just let something else take care of those things, although sometimes when suddenly something stops in front of us.

For us, we see a pedestrian right in front of us, our consciousness comes into play and takes control again, but often we move through the world, not automatically, just not in a way that we are aware of what it is. To say that what we are really seeing is a world where we do not have a unified mind, what we have is a set of different components that make up us and a consciousness that brings those components together and the notion is that for things like memory, we make to memory, to our memories, a question over and over again, different types of questions until we get an answer when we look at things like perception, perception comes to us and gives us information and sometimes we ignore it, so which is sometimes forgotten exactly. where your legs are while you're sitting until it matters and then you become conscious again and then the actions that we take in the world, most of them are automatic, but sometimes we can train and retrain our body and our emotions or our body and our control and what we end up with of these things that look like black boxes that we access and can program, but think about those black boxes that we access in the things that look like devices that are part of us and that we don't fully understand.

I guess they can see everything they see but they don't understand how vision works. They are here but they don't understand how their auditory system works. They can remember but they don't know how their memory works. They can move their head. arms andlegs and they go all over the world, but you don't know how it works, but something in you knows, but your consciousness controls that thing and talks to it. So we have all these things that we do that look like a central unit and all these subsystems with which that interacts, but that's how computer programs work now.

I'm not talking about intelligent programs, I will talk about any program that has its core. and then the subsystems that interact with it and, in particular, when we talk about intelligent systems, we look at things that you have controlled and then these subsystems that are almost always opaque, that's what we call APIs that are opaque to the core. system, so you might think that a machine has a lot of data and analysis, but the reality is that it has these sensors and it doesn't know how they work, but it gets information, it has actors, things that can take action in the world I don't know how they work, but I can tell you what to do combined with things that make decisions and make inferences that understand language, that generate language and now in today's world, machine learning and deep learning use evidence to base their reasoning.

We have to analyze all of these things together in a central repository and therefore the structure of how we build intelligence in a machine is not at all different from the structure that allows intelligence to exist for us and this relationship between a controller and the things that are controlled. the central unit that is aware of what is happening but does not know exactly how each component works is fundamental to intelligence now the reason I care about this the reason I care about this is simple our future is our association with these machines, our future is actually interacting with these machines and to have that future work we have to think about how to humanize them, make them more like us every day because if we don't make them like us, we have to be less like them. humans machines are more mechanized we have to become and so by humanizing them we save ourselves from being mechanized now doing this does not diminish the fact that we can build intelligence and do something that is as if our intelligence does not take anything away from us In fact, on our part It's part of the celebration of being smart that we can turn that gift into a gift for others, a gift for machines, and I always like to go back to Carl Sagan in terms of thinking about this.

Carl Sagan said something beautiful at one point and that is that we are made of star matter, we are made of all the things that are part of the cosmos. I think it's much more important to understand that we are made of the same things as rocks, dirt, mud, and trees. made of the things that are at the basis of creation, but the amazing thing is that even though we are made of those things, we can reason that we can think that we can do this amazing thing and it is not because of the glorious material that we are made of.

It is despite what we are made of and we are now in a place where we can take that wonderful gift of intelligence of reasoning of consciousness of the ability to look at the world and know what to do to decide and think and infer and know it and take it to action. the Machine and that will be for us. I think one of the pinnacles of what it means for us to be intelligent as a species, that's all.