 There we go, um So if it's some point oops if at some point the talks gets a bit weird and you're sitting there like wait I'm a bit lost. I haven't really gotten anything out of it I don't know what's going on That's because I don't know what's going on and nobody really knows what's going on So it's not because you're stupid It's because we have no clue what is happening and it's all open for discussion which makes it really really fun So I will be talking about machine consciousness or rather how we can possibly achieve it And I will conclude with if we do do we really need it and what we're going to do afterwards So point one has anybody seen any of these? How many of you have seen iRobot? How many of you have seen Westworld? How many of you think it's possible to happen? Whoo, exactly. That's the point. Yeah, no, that's not the point Okay Okay, we're back on so many of us think it's possible to happen and In a minute, I will go on to whether we are there yet And how far have we gotten in trying to create machines that are conscious But point one that I want to make is we need to be able to distinguish between a robot and an artificial intelligence So a robot is a machine It's a body and an artificial intelligence is the software that runs on the machine or on your laptop or even on your phone So something to keep in mind in these cases the two kind of work together But for artificial intelligence to work you don't necessarily have to have a body Point one and point two in any case Why would we want to create something that is very powerful and something that is potentially conscious or potentially intelligence? We take inspiration from the human brain and the reason we do that is because the brain is the most powerful computer We actually have nowadays So we know that the brain uses about 15 times less energy than your normal desktop But it can do millions and millions and millions more computations per second than your actual computer can and to put it in a little bit Of numbers to simulate what your brain can do in one second It would take around 80,000 normal computers working for an hour. So it's highly Efficient it's very complex and what we have nowadays doesn't really get there So a lot of computer scientists and a lot of newer scientists put effort into trying to create a computer That's gonna work like a brain because it makes economical and efficient sense So how far have we actually gotten when making conscious and brain-like things in the world? I'm gonna give you three examples which are very different, but very interesting so the first one is a thing called cyber child looks kind of creepy and What it does it's a simulation of a baby It has a bunch of notes and a bunch of things connected to each other And it does what a normal baby does if any of you have a child or have ever communicated with a child with a baby You kind of know that all they do is cry and eat and they go to the toilet And then they cry again and they do it in the cycle until they become a little bit more intelligent I can more explicitly ask for stuff they want and this is literally what cyber child does It has sensors. It has in the virtual world It has sensors inside of it and it can after some time ask for food after it's digested The food is gonna cry to have a diaper changed. So the scientists who created this thing kind of claim that it's conscious Where's it whether it is or whether it's not it's you know up for discussion But that's what they say something to remember is that this is almost 15 years old So we've gotten much further than that but this is often used as an example So here is your basically simulation of a baby. Is it conscious? Is it not up to you to decide? The second one is called Kronos plus Simnes. So Kronos this thing here is an actual robot So it's a physical entity that sits in a lab. It has an eye. It has an arm It's kind of as humane as it could get also almost 15 years ago And with its eye it can see the world around it and there was a virtual simulation of this robot up here called Simnes and what it does in the virtual simulation is Manipulate the objects. This is basically all it does. It can pick them up in the virtual world It can look at them. It can put them down. It can maybe tell you what's on the picture and stuff like that Is it conscious? Is it not again? Probably probably not the interesting thing is that this was created specifically to simulate consciousness So at some point the EU went like alrighty. We need a conscious machine Let's bump millions of dollars into this and they made up this kind of thing and The next one is go. How many of you have heard of go? Probably quite a few. Yeah, it made a big deal Google market in publicity everything like that really really cool Go doesn't really work like a brain does and it's very different from anything I've shown you before It doesn't try to do what a person would do so like ask for food or manipulate its objects But what it can do is do something that we thought only we are able to do so go is an incredibly complex game and The complexity is in the fact that there are so many possible options You can play to win the game that it's mathematically practically impossible to be able to predict Which move will be the winning one so you can you can play something? But you never know whether it's gonna lead you to the win and as people we have learned to feel this game So the the trademark of the game is that you have to feel you have to have a gist for where to play and The fact that a machine was able to actually win this game is a really big deal because it did something It has a feeling it has an intuition for what we thought was only available to us So here are three examples of potentially conscious machines Something that you might have noticed is that they are very very different So point one to remember is that one of them is survival seeking one of them is And the other one is the ability to feel stuff and all of them are very very different And if you dig deeper into it, you will realize that it goes even further so point one everything is different But what even is consciousness so when we try to create a conscious machine We first of all need to decide what do we mean by consciousness and surprise surprise we have absolutely no idea So if you go on Wikipedia, it gives you this nice long definition And it says that consciousness is sentience awareness subjectivity ability to experience and to feel Wakefulness having sense of self who blah blah blah blah blah And if you read papers on philosophy that discuss consciousness that come up with different definitions They talk about behavioral responses introspection ability to suffer and feel pleasure. So they talk about various things Generally, there is really no consensus. So we try to create a machine that does something That we don't even know what it is. So point to take away is We kind of have no idea what we're doing Next point if consciousness does exist, where is it and here? We subscribe to the idea that consciousness happens in the brain for no other reason That if it doesn't happen in the brain Our science is just not gonna work the whole scientific community the whole world We live in all the progress that's made is based on the idea that everything happens in the physical world So there is potential that consciousness is some magic from the sky That there is a God or a deity or some other third force that we don't really know about But if we subscribe to that idea the science is just not gonna work So to make the science work We have to agree that consciousness happens in the brain And this is what we do and we can do cool experiments to try and show where in the brain consciousness happens But the main point is that it happens in the brain and it happens by the work of little cells in the brain That kind of communicate to each other and make the magic. So point three we have to assume it's in the brain Point four if something is conscious How do we judge that it's conscious and here again? There is no really strict definition on this and many scientists have proposed many different theories Some of them say that it's about your communication So some will say that the ability to talk and the ability to think about talking is what makes something conscious Some people say that it's this really complex ability to feel and suffer and experience Emotions and have this super high level cognitive capacity and things like that. Some people say that it's about behaviors But we don't really know what it is and we have this range of hallmarks But we try to judge things if they're conscious or not Well, we kind of have no idea, but does it really matter and Spoiler alert it doesn't and It doesn't matter because of this thing. So this is a very popular philosophical argument called the Chinese room and To explain it very simply is that in this box you have a person who has a dictionary of This is a Chinese character. This is another Chinese character. They go together So this if you get a ball from a person outside with this Chinese character Look it up in the book and give the ball on the outside doesn't make sense So this person inside the box can kind of match the balls the person on the outside Knowing what was the question for example will get the answer and will probably go like Jesus Christ this person is so smart They can actually speak Chinese. They're giving you perfect answers. Do you think this person can speak Chinese? Not really. All they do is just look up things in the book and the real-life example of this We have is Siri. So this is what your Siri does or or the Google live speaking thing in your Android phone that you have You talk to it you ask it a question and gives you the output and sometimes it might be so good that you think god damn it It actually understood what I said not really It just looked it up on the internet basically or it's a known database and it has a really large database So it can do it really really well, but it's not really conscious So we do look up to behavior, but it's kind of useless because of something does behave like it's conscious It doesn't mean that it actually is So doesn't mean we're kind of doomed and we can never create a conscious machine The answer of course is not and there are many outputs and many ways and many Kind of approaches you can take to achieve a conscious machine But I'm gonna focus on one that from the perspective of some you're a scientist is the only foolproof way and Yeah consciousness is a mess. How are you gonna do it? We take the idea that consciousness happens in the brain So we take the idea that it happens by little cells and little points in your brain talking to each other and processing information We combine it with the idea that behavior is not enough So there must be something there not about the behavior But about the processing and about the way the information is Processed and the way this output is produced for something to be conscious We combine these two ideas and boom we have a thing called biologically accurate artificial intelligence And in essence what biologically accurate artificial intelligence is is that it looks at the brain and it tries to do exactly the same Of course, we don't really grow actual cells in an actual laboratory We do a computer simulation of them But what we try to do is have the exact kind of number of cells and the exact types of cells And we try to make them communicate with the dynamics that the brain communicates so that it's exactly as it happens in the brain It doesn't have to be a human brain It can be any animal brain, but it's virtually a copy paste of a biological brain onto a computer So this is quite straightforward How far have we gotten with this with biologically accurate artificial intelligence is a very interesting topic that we can discuss later probably We're not there yet But we have little brain areas that we are able to simulate in a biologically accurate matter So there is potential that one day we can stick them all together and actually end up with the brain So we're gonna try to do it anyway, we're gonna try to do it through biologically accurate Artificial intelligence biologically accurate machines so what then and The big problem here comes is that you have an artificial intelligence You have a machine that basically works exactly the way your brain does But it's a machine So look at it as if it's it's like your child or like your wife or like your best friend It has the same way of thinking it has the same way of feeling it processes information the same way But it doesn't have a physical body that you can hug in the evening. It's a computer How do you treat it? How do you deal with it? What do you do with it? Can you communicate with it? Can you can you put it in prison if it does something wrong? Can you propose to it and get married to it if you really really like it? It's very hard and the difficulty also in this comes that when we talk about artificial intelligence Usually people think about how scary it is So how many of you have ever heard that artificial intelligence is one day gonna take all our jobs and take over the world? Anybody heard that? Yes, some people have I think especially people who work in certain areas It's highly unlikely that things like hairdressers will be done by artificial intelligence But anything that's computational based like engineering or stock market and things like that that can be possibly done by a computer There is fear that one day machine will do that instead and this fear is fueled by people like Stephen Hawking For example who say that developing artificial intelligence one day will put the end basically to everything Which is not necessarily a lie. I mean that might happen But this is exactly as likely as sorry as any of you basically going there and doing something really really bad It's likely as it is for any person But it doesn't mean that this is what's going to happen and most importantly It doesn't mean that this is the only thing that's going to happen So something that I want everybody to take away from the stock and to think about is that if we do create an artificial intelligence that can actually Do exactly the same things as an animal or a human brain can? It will have a moral status and a moral status means that it will have a place in our life and in our world Where we feel like it should be protected Where we feel like it should be granted rights and where we feel like it should be given the opportunity to express itself and be itself And this is where all the difficulties I've mentioned before come. How do you treat it? How do you make it vote in elections? How do you make it reproduces it really wants to reproduce or at one point if it tells you you know, man I don't really want to be in your laptop anymore. I just want to go outside and travel the world because potentially it can How do you make it happen? Kind of for those of you who have seen Westworld This is basically what's gonna what I'm talking about one day is gonna want to run away potentially maybe not maybe yes So this is the point of my talk think about it we can talk about it there is no answer to this question which makes it ever so more interesting and It's about the people it's about what do you think should happen to a machine like that because when a policy is invented People are often asked what do you think you're gonna do with a machine when it has feelings So I'd be really glad to chat to all of you after the talk And answer any of your questions and listen to any of your ideas, but for now, thank you for your attention