Gadgetory

hello my name is Gary Sims from Android authority now we are on the verge of a new era for Humanity and that is the rise of artificial intelligence we already see some of it today in things like Google now and driverless cars over the next few decades we're gonna see lots and lots of changes now Isaac Asimov wrote these three laws of robotics and some people think that actually they're a good set of laws for governing future a I and I'm here to tell you that they're not so let's remind ourselves of the Three Laws of Robotics but before we do that let me tell you about weak AI and strong AI now if you've watched my previous video you'll know there are two types of a I now weak AI is basically a sophisticated computer program that pretends that simulates intelligence so you can ask questions it gives you answers you can even ask about the current state of affairs it will know things you might even ask its favorite color it can tell you but actually it's just a computer program might be neural networks it might be some kind of functions they might be a combination that both is still just a computer program it doesn't actually have self-awareness it can't do abstract thinking it certainly hasn't got free will it's just a computer program that has a very sophisticated user interface now the other type of automated is what they call strong AI and that basically means AI that is self-aware it is a into a mind it's not just a brain it has a mind it can do free thinking you can do abstract thinking it's got free will and basically that's where the difference comes down to it is the free will the ability to choose I like to put it this way if we have a week a iced driverless car you can call it up and say come and get me from the shopping mall and it will just obey is programming and come and get you if you bring up a strong AI and say come and pick me up it could say no I'm watching a movie I don't want to I don't really have the energy right now I'm doing something else it has a free will it is self-aware it is in dependent and those that course are very two different things and that's where the Three Laws of Robotics come into it how can you have something that's going towards intelligence going towards strong AI and yet restrain its actions so it doesn't just do whatever it likes so I'm reading here from a collection of Isaac Asimov robot stories and this is the first story called run around where the three rules are explicitly stated so let's see what it says powers radio voice was terse in Donovan's ear now look let's start with the three fundamental rules of robotics the three rules that are built most deeply into a robot's positronic brain one a robot must not injure a human being or through inaction allow a human being to come to harm - a robot must obey the orders given to it by humans except where such orders would conflict with the first law and three a robot must protect its own existence as long as such protection does not conflict with the first or second laws so there you have it those three laws now before we go any further we must point out that all of Asimov stories are about how those three rules don't work in fact that first story runaround is about how a robot is stuck between laws - and law 3 the law he wanted to obey a command given to it and he wanted to protect its own existence so it was running towards something to fill a command but when it got there it found out was in danger so to start to run back again and in fact it then just started running around in a circle and in fact this is a common problem in computer science in general if you have a thermostat for your room for example and you set it to a certain temperature what you don't want is a room to get to that temperature and then to drop down by even 0.1 of a degree and then the heating come back on again and then it would go back to the temperature and then it would switch off and then it'd go down and you would basically get your heating go on off on off on off on off all the time so what you actually do is you build in a range you say when it gets up to this then it dropped down by this then go back up again and in fact that's a simple thing in computer science but what azimoff is showing here is that when you state things explicitly like this they can be easily misinterpreted so let's look at rule number one for example no harm should come to a human being well that's easy when you're looking at the ideas of a car is about to crash into somebody and robot dashes across the road grabs you and drags you out a way of the car that's great but what does harm mean I mean physical harm emotional harm psychological harm I mean there's so many different ways to define the word harm now in the books actually Asimov goes to the logical conclusion that actually robots are allowed to lie because if they tell someone the truth they could actually be harming them emotionally so obviously that doesn't work all right things like smoking or fast food smoking is considered across the world to be dangerous harmful to your health and yet there are millions of people that choose to smoke and therefore if this rule was implemented literally robots would have nothing to do other than to go and pull cigarettes out of people's mouths because they say they were harming them or what about fast food if you eat too much fast food too much saturated fats too many bad ingredients it's gonna do you harm so do robots March around closing down rest fast food restaurants I mean what does harm mean is that such an ambiguous definition that it's no good for the context of defining the behavior of a robot and ultimately Asimov wrote a story where he showed that harm could be interpreted as not harm to a human but harm to humanity and therefore you have the rise of the machines they try to take over actually to protect humanity from our own errors and from our own bad actions and what about this idea it has to obey a rule from a human well which human mean is a three-year-old the same as a ten year old is a ten year old the same as a thirty year old is a normal citizen the same as a police officer is a police officer the same as a member of a parliament you could just keep defining roles again and again and again so if a three year old says to a robot let's jump up and down on the sofa actually a robots probably gonna be pretty heavy he's gonna wreck that sofa pretty quickly but however it was obeying a command and no humans came to harm of course when mum and dad come home they'll say what are you doing and it will immediately stop but it those rules don't define what it should listen to a three-year-old or not the problem with the three rules is there's no moral compass it's got no way of knowing what's right and what's wrong so for example let's say you'd have a driverless car you send out go and get me some fast food and it drives to the fast food place and you find out it's closed so it looks on its map and it says right there's a pizza place 50 miles away and so off it goes to the 50 mile away or Peter now that's not the right thing to do obviously now at some point it maybe would need to have further instructions from its owners but how does it know the difference between what's good and what's bad the three rules don't tell it that now maybe going an extra 50 miles for a pizza wouldn't be a good idea but you've heard a sick child and you were desperate to get some medication and the nearest open available source that medication was 50 miles away you'd probably want the robot to go get it but how does it choose those three rules don't tell it anything they don't help define what's right and what's wrong now we as humans we look at what's right and what's wrong every single day and sometimes we get it right and sometimes we get it wrong and there are debates even at a national level about what is right and what is wrong what is moral and what is immoral now how does a robot get that does it get that by looking at a data set of people's actions does it study what humans do and then define what is right and what is wrong because actually if you look at what we do as humanity then actually there are some pretty terrible things that we do and you wouldn't want a robot to take over from those things to learn from those things so then there's the question of Cameron ITB learned from a data set and what about the people who are programming the robots do they program their morality into it rather than something that another group of people would consider to be right or because it to be wrong and what about the idea that we all want to be better than we really are we all want to be different but yet we find there's a weakness in us that means we can't always do the things we want to do and sure you wouldn't want a robot to have that you'd want the robot to be a perfect moral character where does it get morality certainly doesn't get them from the Three Laws of Robotics so as we're defining the progress of artificial intelligence we are start to need to ask questions about freewill and about morality because those things will define how a robot behaves not whether it should just pull a person out from in front of a car or not well my name's go Sims from Andrew Authority I hope you enjoyed this video and look at the future now why we're looking at this is because we are on the verge of got self-driving as cars we've got google's AI already doing so many different things on means I even read there are now recruitment agencies that are using AI to scan through CV so if your CV doesn't match up to what the AI is expecting you don't even get called for an interview we know that Google search results are controlled in a certain way and I'm sure there are different types of weak AI involved in that searching and how that is filtered for different people for different reasons regions for different types of search results we're going to be finding AI creeping into so many different things and the question is where does it get his guidance about what's right and what's wrong also you want to talk more about this poke to the annual Authority for oh there's a special topic that I've opened there just so that we can discuss the three laws of robotics I look forward to seeing you there well if you enjoyed this video please do give it a thumbs up don't forget to subscribe to and royalties YouTube channel you can follow me on Twitter and on Instagram and on Google+ and you can follow handwritten thority the same on all those social media networks don't forget to download the Android authority app and don't forget to go to andrew authority comm because we are your source for all things Android