WEBVTT - Robots are People Too? 0:00:04.120 --> 0:00:07.160 Get in touch with technology with tech Stuff from how 0:00:07.200 --> 0:00:13.880 stuff Works dot com. Hey there, and welcome to tech Stuff. 0:00:13.880 --> 0:00:17.959 I'm your host, Jonathan Strickland. I'm an executive producer and 0:00:18.079 --> 0:00:21.560 I love all things tech. And today we're gonna talk 0:00:21.600 --> 0:00:28.440 about the concept of granting personhood or personality two robots, 0:00:28.480 --> 0:00:31.280 not a personality in the sense of this robot is 0:00:31.360 --> 0:00:35.640 Chipper and this one's depressed, but rather the concept of 0:00:35.880 --> 0:00:42.800 granting robots some of the aspects that we say humans 0:00:43.000 --> 0:00:48.720 have from a legal standpoint, and there are numerous arguments 0:00:48.760 --> 0:00:54.800 both for and against this concept. Uh. I've talked about 0:00:54.960 --> 0:00:59.000 artificial intelligence extensively, not just on this show but on 0:00:59.120 --> 0:01:03.040 other shows as well. And one of the elements about 0:01:03.120 --> 0:01:07.319 artificial intelligence that tends to pop up, especially in things 0:01:07.319 --> 0:01:12.440 like science fiction, is what happens when artificial intelligence is 0:01:12.480 --> 0:01:19.040 able to take actions that could negatively impact people? Uh, 0:01:19.120 --> 0:01:21.520 what do we do? What sort of framework do we 0:01:21.600 --> 0:01:26.160 make if that were the case, where we can determine 0:01:26.240 --> 0:01:29.319 who's responsible for this? And then the interesting thing is 0:01:29.360 --> 0:01:34.759 that for me anyway, the European Parliament, a group out 0:01:34.760 --> 0:01:37.440 of the European Parliament kind of a working group, put 0:01:37.480 --> 0:01:41.800 together a proposal a couple of years ago about the 0:01:41.840 --> 0:01:45.240 possibility of this and the idea of granting personhood to 0:01:45.440 --> 0:01:49.400 robots in a way to create an established framework of 0:01:49.480 --> 0:01:54.760 law and accountability, because at the moment there's not really 0:01:54.800 --> 0:01:58.800 anything formalized there, so they put together a proposal. The 0:01:58.840 --> 0:02:04.520 proposal UH is still online. It was originally published on 0:02:04.720 --> 0:02:08.000 May thirty one, two thousand sixteen. It was submitted in 0:02:08.080 --> 0:02:11.720 draft form. The title of the draft was Draft Report 0:02:11.800 --> 0:02:15.840 with Recommendations to the Commission on Civil Law Rules on 0:02:16.080 --> 0:02:20.560 Robotics and it's a fascinating report. It's actually a lot 0:02:20.560 --> 0:02:22.799 of fun to read. I highly recommend checking it out 0:02:22.880 --> 0:02:26.240 if you have some time. The general purpose of the 0:02:26.280 --> 0:02:31.280 proposal was to start official discussions in the European Union 0:02:31.760 --> 0:02:36.040 about developing policies and guidelines in the field of robotics, 0:02:36.080 --> 0:02:41.639 particularly since there were individual member states of the EU 0:02:41.760 --> 0:02:45.799 s several nations that were developing their own policies over time, 0:02:46.639 --> 0:02:51.080 and the European Parliament Working Group was saying this could 0:02:51.080 --> 0:02:54.920 be a problem because if we have one set of 0:02:54.960 --> 0:02:59.440 policies that are in play and let's say France, and 0:02:59.480 --> 0:03:03.040 a different set that are in play in Germany, that 0:03:03.560 --> 0:03:07.600 starts to create conflict, and in the European Union, you're 0:03:07.600 --> 0:03:10.280 supposed to be able to move throughout the union and 0:03:10.320 --> 0:03:14.799 take jobs wherever and live wherever within the Union. And 0:03:15.360 --> 0:03:19.120 therefore it would work a lot better if everyone had 0:03:19.160 --> 0:03:22.960 the same set of rules and regulations on something like this, 0:03:23.080 --> 0:03:28.200 especially something that's going to factor so heavily in industry 0:03:28.360 --> 0:03:33.480 and economy in the workplace. So really this was mostly 0:03:33.520 --> 0:03:36.560 a kind of a warning saying we need to start 0:03:36.560 --> 0:03:40.760 talking about this. The introduction of that proposal is phenomenal 0:03:40.800 --> 0:03:43.960 in its own right. It cites lots of literary works, 0:03:44.000 --> 0:03:50.480 including Mary Shelley for Frankenstein, The Pygmalion, myth Uh. It 0:03:51.120 --> 0:03:56.920 references Carol Capex Are You Are? That is the play 0:03:56.960 --> 0:04:03.520 that originated the word robot. So it's pretty interesting just 0:04:03.680 --> 0:04:07.760 from the fact that it's referencing science fiction and horror 0:04:08.320 --> 0:04:12.320 literature way more than you would expect in your typical 0:04:13.000 --> 0:04:17.480 government proposal. It's it's much more entertaining than reading a 0:04:18.279 --> 0:04:22.560 draft of, say, uh like your typical agricultural report, which 0:04:22.560 --> 0:04:25.520 is still incredibly important. Don't get me wrong, those are 0:04:25.720 --> 0:04:29.960 very important reports. They just don't tend to be page turners. Now, 0:04:29.960 --> 0:04:33.240 those citations are all there to establish that we humans 0:04:33.279 --> 0:04:38.840 have been really fascinated, maybe even fixated with this idea 0:04:39.400 --> 0:04:43.560 of creating intelligent machines or even intelligent life in the 0:04:43.600 --> 0:04:47.520 case of Frankenstein's Monster, that this has been something that 0:04:47.600 --> 0:04:52.599 we have really aspired to on some level, the idea 0:04:52.640 --> 0:04:56.320 that we create something that itself could be said to 0:04:56.440 --> 0:05:00.960 be intelligent. Section B of the introduction, as the introductions 0:05:01.000 --> 0:05:04.400 long enough to have different sections, would say that we're 0:05:04.440 --> 0:05:09.279 on the threshold of a new industrial revolution. That the 0:05:09.320 --> 0:05:13.480 first industrial revolution that took place in the nineteenth century 0:05:13.800 --> 0:05:20.360 was all about creating factories, harnessing the power of coal building, railroads, 0:05:20.960 --> 0:05:25.960 steam power, assembly lines, all of that stuff transformed the 0:05:26.000 --> 0:05:28.760 world from what what had been a really a grarian 0:05:28.960 --> 0:05:33.240 society into more of an urban one in these industrialized nations. 0:05:34.240 --> 0:05:38.960 And the section was saying that artificial intelligence was going 0:05:39.000 --> 0:05:42.520 to fuel a new revolution that would be just as 0:05:42.600 --> 0:05:46.880 impactful as the previous industrial revolution, that no part of 0:05:46.960 --> 0:05:51.239 society would go unaffected by that revolution. So therefore, since 0:05:51.240 --> 0:05:54.599 it was going to be such a big part of 0:05:54.600 --> 0:05:57.440 our lives moving forward, it would be really smart for 0:05:57.520 --> 0:06:00.839 us to think about the implications of that out before 0:06:00.880 --> 0:06:04.359 it happens, and prepare for it and to put into 0:06:04.440 --> 0:06:09.520 place protections for people before it becomes reactionary, you know, 0:06:09.600 --> 0:06:11.960 before there's a problem and then we have to figure out, oh, 0:06:12.000 --> 0:06:15.359 how can we fix this. The argument of the proposal was, 0:06:15.560 --> 0:06:19.440 let's be proactive, let's try and figure out what problems 0:06:19.520 --> 0:06:24.200 may confront us in the future and solve them now 0:06:24.800 --> 0:06:31.039 before we are dealing with a catastrophic experience. So then 0:06:31.080 --> 0:06:34.240 the proposal goes and points out that robots sales have 0:06:34.320 --> 0:06:36.640 been on the rise over the past few years. They 0:06:36.760 --> 0:06:40.480 they've increased year over year. The automotive industry in particular 0:06:40.880 --> 0:06:44.080 was called out, and it also stated that robots will 0:06:44.120 --> 0:06:48.120 provide numerous benefits in the short to medium term and 0:06:48.160 --> 0:06:54.040 potentially what they called virtually unbounded prosperity in the long term. 0:06:54.120 --> 0:06:58.080 So the idea being there are some real positive outcomes 0:06:58.200 --> 0:07:02.640 to employing robots. However, the flip side of that is 0:07:02.680 --> 0:07:06.600 those advances might quote result in a large part of 0:07:06.600 --> 0:07:09.240 the work now done by humans being taken over by 0:07:09.320 --> 0:07:12.760 robots end quote, and that that would affect not just employment, 0:07:12.840 --> 0:07:16.720 but systems like social security which rely upon employment taxes 0:07:16.720 --> 0:07:20.600 for funding and other things that are tax supported. And 0:07:20.680 --> 0:07:22.840 what happens when a robot fails in some way? What 0:07:22.880 --> 0:07:26.360 if it creates damage to a person or property? And 0:07:26.640 --> 0:07:30.320 how will robots interact with our personal data, even you know, 0:07:30.600 --> 0:07:34.400 personal data that we haven't necessarily given consent to share. 0:07:35.160 --> 0:07:38.200 The more intelligent the system, the more proactive it might 0:07:38.240 --> 0:07:41.640 be in taking that data and using it in some 0:07:41.720 --> 0:07:45.120 way without us even saying anything about it, especially if 0:07:45.160 --> 0:07:47.680 you haven't built in specific rules for the robot or 0:07:47.760 --> 0:07:51.240 artificial intelligent program or whatever to follow so that it 0:07:51.320 --> 0:07:55.000 doesn't do that. Now, before I go much further down 0:07:55.080 --> 0:07:58.160 that particular line of reasoning, I do need to point 0:07:58.160 --> 0:08:03.080 out that AI and automation and their effect on jobs 0:08:03.360 --> 0:08:07.520 is still a matter of considerable debate. No one is 0:08:07.560 --> 0:08:10.480 sure right now to what extent it's going to have 0:08:10.600 --> 0:08:14.160 an effect on people in general. So the worst case 0:08:14.160 --> 0:08:18.160 scenario that some people say is that robots and AI 0:08:18.280 --> 0:08:21.760 automated systems are going to replace the vast majority of 0:08:21.880 --> 0:08:25.960 jobs within a few decades, and that this will happen 0:08:26.080 --> 0:08:29.440 before we've ever built out any sort of infrastructure that 0:08:29.480 --> 0:08:33.040 would take care of people that would help them transition 0:08:33.440 --> 0:08:37.280 either into new jobs that were created as a result 0:08:37.400 --> 0:08:40.839 of this, jobs that we don't have today because they 0:08:41.320 --> 0:08:44.000 we don't need them, but we may have a need 0:08:44.040 --> 0:08:48.480 of them once this automated future comes upon us. Or 0:08:49.600 --> 0:08:52.760 maybe that isn't the case. Maybe we don't create new jobs, 0:08:53.080 --> 0:08:56.800 Maybe we don't have a system where we can actually 0:08:57.200 --> 0:09:01.800 take care of people and and separate the concepts of 0:09:02.080 --> 0:09:06.960 making a living and being employed, having some other system 0:09:07.040 --> 0:09:10.720 there so that people could meet their needs to make 0:09:10.880 --> 0:09:14.080 ends meet. How do we do that without having it 0:09:14.120 --> 0:09:18.440 tied to jobs? But there are other people who also 0:09:18.480 --> 0:09:22.160 think we're not likely to see that huge of a change, 0:09:22.200 --> 0:09:27.040 at least not in the relatively near future. We're more 0:09:27.160 --> 0:09:31.240 likely going to see AI and automation take over individual tasks, 0:09:31.720 --> 0:09:35.240 but not entire jobs. So the future might be one 0:09:35.280 --> 0:09:38.360 in which our work is augmented by AI, but we 0:09:38.440 --> 0:09:43.200 are not replaced by AI. The justification for that argument 0:09:43.280 --> 0:09:46.520 is saying, look at machine learning as it stands right now. 0:09:46.559 --> 0:09:50.080 It is very impressive, but it also shows that there 0:09:50.080 --> 0:09:54.360 are severe limitations in machine learning right now. It's just 0:09:54.440 --> 0:09:58.480 not sophisticated enough to be able to take over for 0:09:58.600 --> 0:10:01.760 everything that a human can do. Uh, and it requires 0:10:01.800 --> 0:10:03.440 a lot of training and a lot of stuff can 0:10:03.480 --> 0:10:07.200 go wrong, So it's not likely that we're gonna see 0:10:07.200 --> 0:10:10.240 computers take over lots and lots and lots of jobs, 0:10:10.240 --> 0:10:13.280 but they might take over repetitive tasks that your job 0:10:13.480 --> 0:10:16.319 happens to include, and then you can focus on the 0:10:16.360 --> 0:10:21.120 stuff that isn't repetitive and predictable. Who's to say, who's 0:10:21.200 --> 0:10:24.440 right not me? I don't know. It could be either 0:10:24.480 --> 0:10:26.280 one of those. But the argument of the proposal I 0:10:26.320 --> 0:10:28.840 think still holds true, which is that we need to 0:10:29.559 --> 0:10:33.160 at least consider what the worst case scenario is and 0:10:33.440 --> 0:10:38.880 have a plan to alleviate the outcome of that. Then 0:10:38.920 --> 0:10:41.520 there's the question of how robots are going to impact 0:10:41.720 --> 0:10:45.120 stuff like human dignity. So in the future, if we 0:10:45.200 --> 0:10:48.559 have robots that are acting like caretakers for the sick 0:10:48.960 --> 0:10:51.600 or the elderly or the young, what impact is that 0:10:51.640 --> 0:10:54.920 going to have on those people, How is that going 0:10:54.960 --> 0:10:57.880 to make them feel? What impact will have on their health? 0:10:58.679 --> 0:11:01.200 And uh, and hey, what if those robots were to 0:11:01.240 --> 0:11:07.320 get really effective, really smart, like maybe smarter than humans smart, 0:11:07.840 --> 0:11:12.199 So again, not necessarily intelligent in the same way humans are, 0:11:12.240 --> 0:11:15.199 but better able to process certain types of information than 0:11:15.400 --> 0:11:20.720 humans are. The proposal in the introduction, it raises that 0:11:20.800 --> 0:11:23.800 question as well. Could robots actually represent a danger to 0:11:23.840 --> 0:11:26.840 the human species. It's something that we have to consider 0:11:27.400 --> 0:11:31.360 before it becomes a reality. Now, with all of this 0:11:31.480 --> 0:11:35.360 in mind, says the introduction, the EU should get off 0:11:35.360 --> 0:11:38.880 its butt and start talking about those ideas and work 0:11:38.880 --> 0:11:42.760 on a strategy to avoid problems in the future. And 0:11:43.280 --> 0:11:46.320 the proposal would then go on to have a few 0:11:46.360 --> 0:11:48.480 suggestions of its own. I'll get into those in just 0:11:48.600 --> 0:11:51.760 a second, but first let's take a quick break to 0:11:51.880 --> 0:12:03.480 thank our sponsor. So the proposal has a solutions and 0:12:03.559 --> 0:12:08.760 Suggestions section, and under general principles, the report would cite 0:12:08.840 --> 0:12:14.600 Asimov's Laws of Robotics and would say that designers, producers, 0:12:14.600 --> 0:12:18.600 and operators of robots should keep these in mind, which 0:12:18.679 --> 0:12:22.720 is amusing to me. I mean asthma asthmava. Isaac Asimov 0:12:23.000 --> 0:12:29.120 was a famed science fiction and speculative fiction author wrote 0:12:29.160 --> 0:12:33.160 some amazing stuff, and those laws of robotics have become 0:12:33.200 --> 0:12:38.680 sort of iconic in artificial intelligence. The original three laws 0:12:38.720 --> 0:12:41.959 of robotics are probably the best known. So law number 0:12:42.000 --> 0:12:44.880 one is a robot may not harm a human or 0:12:45.120 --> 0:12:48.599 through in action, allow a human to come to harm. 0:12:48.840 --> 0:12:52.439 Law too is a robot has to obey any order 0:12:52.800 --> 0:12:56.480 given to it by a human unless it would violate 0:12:56.720 --> 0:13:00.000 the first law. So you could tell a robot, hey 0:13:00.000 --> 0:13:01.440 will pick that up for me, and it would have 0:13:01.520 --> 0:13:04.400 to do it. But you couldn't tell robot, hey, go 0:13:04.600 --> 0:13:07.360 push that guy into traffic, because that would violate the 0:13:07.400 --> 0:13:10.600 first law. The third law was a robot would have 0:13:10.679 --> 0:13:15.280 to protect itself from harm unless doing so would conflict 0:13:15.320 --> 0:13:19.400 with either of the first two laws. So if a 0:13:19.559 --> 0:13:23.600 robot were to see that a car were coming, it 0:13:23.640 --> 0:13:25.439 was coming down the street, it was gonna hit a 0:13:25.480 --> 0:13:27.920 little old lady, and the robot would be able to 0:13:27.960 --> 0:13:30.559 push the little lady out of the way, but as 0:13:30.559 --> 0:13:33.120 a result, it was going to get hit by this car. 0:13:33.320 --> 0:13:35.400 The robot would have to do it because even though 0:13:35.440 --> 0:13:38.240 it has a law stating it has to protect itself, 0:13:38.760 --> 0:13:41.960 it that gets superseded by law number one, which says 0:13:42.480 --> 0:13:44.920 it cannot through in action, allow a human to come 0:13:44.960 --> 0:13:47.160 to harm. It would have to take action in that case. 0:13:47.880 --> 0:13:51.520 Then there's a fourth law called law zero that was 0:13:51.559 --> 0:13:55.360 considered to be the top of the this ladder, the 0:13:55.480 --> 0:13:57.920 most important of all the laws, which is a robot 0:13:57.960 --> 0:14:02.280 may not harm humanity or by inaction allow humanity to 0:14:02.320 --> 0:14:04.960 come to harm, so not just an individual, but humanity 0:14:05.080 --> 0:14:09.880 as a whole. Under the liability section of the report, 0:14:10.280 --> 0:14:12.960 it suggests that it will not be long before the 0:14:13.000 --> 0:14:17.000 European Union either needs to classify robots under a category 0:14:17.040 --> 0:14:20.400 such as persons or to create a brand new category 0:14:20.520 --> 0:14:25.240 just for robots. So AI and machine learning are really 0:14:25.640 --> 0:14:31.880 changing how robots interact with environments. Before you would program 0:14:31.920 --> 0:14:35.960 all the ways that a robot would interact with its environment, 0:14:36.080 --> 0:14:40.160 and mostly you would try to control the types of 0:14:40.240 --> 0:14:42.680 environments your robot was going to be in. You know, 0:14:42.760 --> 0:14:46.120 the more predictable and the more stationary the robot, the 0:14:46.160 --> 0:14:48.240 easier it was to program. Right, So if you've got 0:14:48.240 --> 0:14:51.280 a robot, like a giant robotic arm that's doing welding 0:14:51.400 --> 0:14:54.640 in a car manufacturing line, that robot is going to 0:14:54.680 --> 0:14:58.080 be stationary. It doesn't move around. It stays in one 0:14:58.120 --> 0:15:00.840 spot along the assembly line. The car has come to it, 0:15:00.840 --> 0:15:05.200 it does its work, next car comes to it, it continues. Uh. 0:15:05.280 --> 0:15:09.600 That is one way of programming a robot, and it's 0:15:09.800 --> 0:15:14.440 easy comparatively speaking. But these days we now have robots 0:15:14.480 --> 0:15:19.960 that use machine learning that encounter situations and then process 0:15:20.000 --> 0:15:24.320 information come up with a conclusion. It might be a 0:15:24.320 --> 0:15:28.520 way to act, might be UH, you know, a specific 0:15:29.200 --> 0:15:33.320 um request, it might make whatever you may think, it 0:15:33.400 --> 0:15:38.480 has to come to that conclusion, which means that we 0:15:38.560 --> 0:15:41.880 don't know how robots are always going to react to 0:15:41.920 --> 0:15:46.320 their environments. Environments can be very chaotic things with lots 0:15:46.320 --> 0:15:49.440 of different variables. And while you might program a robot 0:15:49.480 --> 0:15:52.120 so it will behave in a very predictable way for 0:15:52.240 --> 0:15:55.720 certain situations, you're not gonna be able to predict every 0:15:55.760 --> 0:16:00.320 possible situation situation in every possible environment. So because of that, 0:16:00.440 --> 0:16:05.200 and because we're using machine learning in more applications, this 0:16:05.240 --> 0:16:08.600 program was saying, or this this UH report was saying, 0:16:09.360 --> 0:16:12.640 we need to keep that in mind. And while we 0:16:13.480 --> 0:16:17.240 wouldn't necessarily call even the most advanced AI conscious or 0:16:17.240 --> 0:16:20.560 self aware, it's not. It's not doesn't appear to be. 0:16:21.200 --> 0:16:23.840 We are seeing more applications that allow machines to learn 0:16:23.840 --> 0:16:27.440 from their environments and adapt their approaches to complete certain tasks, 0:16:27.560 --> 0:16:29.320 and that is something we need to keep an eye 0:16:29.360 --> 0:16:31.920 on now. Right now, you could argue that if a 0:16:32.040 --> 0:16:37.040 robot malfunctions, you would say the company that made the 0:16:37.120 --> 0:16:41.320 robot should be held liable or the programmer who programmed 0:16:41.360 --> 0:16:45.880 in either the routine the robot was following or the 0:16:45.880 --> 0:16:51.280 basic artificial intelligence that the robot would use as guideline. However, 0:16:51.440 --> 0:16:54.360 as robots depend more and more upon machine learning to 0:16:54.440 --> 0:16:58.640 interact with their environments, this gets really murky because again, 0:16:58.680 --> 0:17:02.680 those environments are filled with predictable variables, and the robot 0:17:02.800 --> 0:17:05.280 might quote unquote learned to do something in a way 0:17:05.320 --> 0:17:09.680 that's harmful, or it's inefficient, or it's ineffective. This doesn't 0:17:09.720 --> 0:17:12.160 necessarily mean it would cause harm to people. It might 0:17:12.280 --> 0:17:16.000 just mess things up in like an industry environment and 0:17:16.119 --> 0:17:21.760 cause financial harm. So the proposal argues that as machines 0:17:21.840 --> 0:17:27.640 become more advanced, it makes less sense to blame the manufacturer. 0:17:27.720 --> 0:17:30.880 It's less effective, and there might need to be new 0:17:30.960 --> 0:17:34.680 rules and definitions for liability that would hold the machines 0:17:35.000 --> 0:17:41.399 themselves responsible. So the proposal quote considers that a system 0:17:41.440 --> 0:17:45.399 of registration of advanced robots should be introduced end quote. 0:17:45.640 --> 0:17:49.400 That would involve creating the criteria to decide which robots 0:17:49.440 --> 0:17:52.720 would be required for registration, and it would also call 0:17:52.760 --> 0:17:56.200 for more funding from the European Union for research projects, 0:17:56.440 --> 0:18:01.080 particularly those that involve the social and ethical challenges raised 0:18:01.119 --> 0:18:04.639 by advances in robotics, and that the EU should create 0:18:04.720 --> 0:18:09.000 a quote Legislative Instrument on legal questions related to the 0:18:09.000 --> 0:18:12.240 development of robotics and AI end quote that would look 0:18:12.280 --> 0:18:16.560 ahead ten to fifteen years, which is super hard to do. 0:18:17.200 --> 0:18:20.680 It's really hard to predict what will happen in technology 0:18:20.720 --> 0:18:24.480 in five years, let alone ten or fifteen. But they're 0:18:24.480 --> 0:18:27.840 saying you should try and consider as many different possible 0:18:27.880 --> 0:18:33.320 situations as you can. So why are they saying this, Well, 0:18:33.840 --> 0:18:38.320 if I create a machine learning program and I have 0:18:38.400 --> 0:18:42.280 a robot that follows it, and the robot is put 0:18:42.320 --> 0:18:45.200 into an environment that's got a lot of these variables, 0:18:46.000 --> 0:18:48.000 I may not know how my robots going to react 0:18:48.000 --> 0:18:51.760 in every single situation. The more advanced the robot is, 0:18:52.480 --> 0:18:55.120 the less certain I can be of exactly what it's 0:18:55.119 --> 0:19:01.120 going to do based upon any given scenario. I might 0:19:01.480 --> 0:19:04.640 feel like the guidelines I've created are enough to keep 0:19:04.680 --> 0:19:07.280 the robot out of trouble. But the world is a 0:19:07.359 --> 0:19:10.800 chaotic place, and we've talked about this with autonomous cars. 0:19:10.840 --> 0:19:12.600 I'm actually going to talk about it more. I plan 0:19:12.680 --> 0:19:16.520 on doing a whole suite of episodes about the history 0:19:17.320 --> 0:19:21.960 an evolution and challenges of autonomous cars in the near future. 0:19:22.000 --> 0:19:27.600 That's gonna be a multi part series. But with autonomous cars, 0:19:28.280 --> 0:19:34.480 we know that there are situations that can end in accidents, 0:19:34.600 --> 0:19:39.240 even fatal ones. We've seen examples of that, and that 0:19:39.760 --> 0:19:45.240 this raises questions of what ultimately is what should we 0:19:45.280 --> 0:19:51.200 hold accountable, who who is responsible for this? With driving scenarios, 0:19:51.640 --> 0:19:55.800 we know we can predict maybe let's say nine of 0:19:55.840 --> 0:19:58.520 the different scenarios you would encounter on a typical drive, 0:19:58.920 --> 0:20:00.680 but that means there's ten percent and the stuff out 0:20:00.720 --> 0:20:04.200 there that happens, it's just not normal. It's not something 0:20:04.200 --> 0:20:09.000 you would typically encounter. And these outside scenarios pose a 0:20:09.040 --> 0:20:12.840 problem because you cannot anticipate every single one in program 0:20:12.960 --> 0:20:16.920 into your machine when this happens, make sure you do 0:20:17.000 --> 0:20:21.080 this other thing. The machine will encounter scenarios in which 0:20:21.119 --> 0:20:24.320 it has to make a decision for itself, and it's 0:20:24.359 --> 0:20:28.959 at that point that we don't know where responsibility falls, 0:20:29.040 --> 0:20:34.080 where the accountability falls because the programmer could not possibly 0:20:34.119 --> 0:20:37.400 have anticipated this, So is it really fair to hold 0:20:37.440 --> 0:20:41.480 them at fault. The manufacturer may have made it exactly 0:20:41.480 --> 0:20:43.359 the way it was supposed to be made. There are 0:20:43.359 --> 0:20:46.360 no faults in the manufacturing process, So are they at 0:20:46.359 --> 0:20:50.240 fault the owner of the robot that put it in 0:20:50.640 --> 0:20:54.320 whatever situation it found itself in the first place, maybe 0:20:54.359 --> 0:20:57.959 they're at fault, or maybe the robot itself is at fault. 0:20:58.880 --> 0:21:02.679 So this proposal goes on to call for a legal 0:21:02.720 --> 0:21:07.120 solution that doesn't restrict the type or extent of damage 0:21:07.160 --> 0:21:10.080 as a person can seek based solely on the fact 0:21:10.160 --> 0:21:13.120 that the damage was caused by a non human agent. So, 0:21:13.359 --> 0:21:16.040 in other words, a court should should not be allowed 0:21:16.040 --> 0:21:19.600 to say, well, we can only award you this amount. 0:21:19.680 --> 0:21:22.960 We can't give you any more than this because the 0:21:23.000 --> 0:21:25.760 thing that caused harm to you was a robot, not 0:21:25.840 --> 0:21:27.920 a person. If it were a person, we would be 0:21:27.960 --> 0:21:30.760 able to award you more money. The proposal says, we 0:21:31.040 --> 0:21:33.280 want to make sure that does not become the case. 0:21:33.800 --> 0:21:36.960 And like I said, we're already seeing this in cases 0:21:36.960 --> 0:21:41.600 with autonomous cars, where a company can say, well, you 0:21:41.640 --> 0:21:45.760 know it was it was not the fault of the programmer, 0:21:45.880 --> 0:21:47.960 was not the fault of the company. It was a 0:21:48.000 --> 0:21:51.000 situation in which the car itself made that decision. The 0:21:51.080 --> 0:21:55.440 car is at fault, not us. That is a problematic thing. 0:21:56.040 --> 0:21:59.639 One of the things that the the report suggested was 0:22:00.119 --> 0:22:02.840 that the producers of a robot are liable for damage 0:22:02.840 --> 0:22:05.439 on a level that is proportionate with the amount of 0:22:05.480 --> 0:22:09.680 instructions the producers gave to the robot. So the more 0:22:10.160 --> 0:22:13.920 locked down the robot, the more responsible the producers are 0:22:14.000 --> 0:22:17.600 for that robot's behavior. So if you have that stationary 0:22:17.880 --> 0:22:21.840 robot that that is welding on an assembly line and 0:22:21.960 --> 0:22:25.879 it malfunctions or it starts causing huge amounts of damage 0:22:25.880 --> 0:22:30.960 and its behavior it's it's perhaps ruining several cars along 0:22:31.000 --> 0:22:33.800 the assembly line as it's going through this process, you 0:22:33.840 --> 0:22:36.280 would say, all right, the company that made this robot 0:22:36.400 --> 0:22:39.560 is at fault. Something they did was wrong because the 0:22:39.640 --> 0:22:43.760 robots only following the instructions that this company gave it. 0:22:43.760 --> 0:22:46.359 It's not making any decisions on its own. But the 0:22:46.400 --> 0:22:50.439 more the decision making process is on the robot, the 0:22:50.560 --> 0:22:54.639 less you can hold the manufacturer accountable, according to this report, 0:22:54.640 --> 0:22:58.000 which is a pretty radical idea, and the proposal also 0:22:58.040 --> 0:23:01.760 states that the longer a robot has received quote unquote education, 0:23:02.160 --> 0:23:05.639 the more liable the robots quote unquote teacher is for 0:23:05.680 --> 0:23:08.720 any damage the robot causes. Now, that might be the 0:23:08.760 --> 0:23:10.680 company that made the robot. It might be the person 0:23:10.760 --> 0:23:13.440 that programmed the robot, it might be the person who 0:23:13.640 --> 0:23:16.920 purchased the robot and then put it in an environment 0:23:16.960 --> 0:23:20.359 where it was learning, but it would be whoever was 0:23:20.440 --> 0:23:25.359 introducing these scenarios to the robot. So that's a really 0:23:26.040 --> 0:23:28.880 interesting concept. And again, this is just a proposal. It's 0:23:28.880 --> 0:23:32.440 not like this has been enacted into law. Uh. They 0:23:32.520 --> 0:23:36.960 also created a suggestion for an obligatory insurance scheme. This 0:23:37.080 --> 0:23:39.800 is not that crazy. I mean, if you drive a 0:23:39.840 --> 0:23:42.600 car in the United States, you have to have insurance. 0:23:43.240 --> 0:23:45.639 It's a requirement by law. So it'd be similar to that, 0:23:45.680 --> 0:23:48.920 except in this case, the producers of the robots would 0:23:48.920 --> 0:23:52.120 pay out the insurance for the robots that it was creating. 0:23:52.640 --> 0:23:55.920 And the report also suggested that there should be or 0:23:55.960 --> 0:24:01.080 there could be, a compensation fund for the robots, which 0:24:01.119 --> 0:24:03.880 sounds crazy. Like one of the things you think about 0:24:03.960 --> 0:24:07.199 about automation is that, well, it reduces the need to 0:24:07.280 --> 0:24:11.080 have paid employees. You've got robots, why do you need 0:24:11.119 --> 0:24:15.040 to pay them. Well, it's not to reward the robot. 0:24:15.080 --> 0:24:17.480 It's not to make the robot feel like it did 0:24:17.480 --> 0:24:21.359 a great job. Robots can't feel anything anyway. It's not 0:24:21.560 --> 0:24:26.520 even to you know, to to have this system, you know, 0:24:26.680 --> 0:24:29.960 perpetuate the system of paying for work. It's rather to 0:24:30.160 --> 0:24:34.119 build a fund, a compensation fund that could cover the 0:24:34.160 --> 0:24:37.840 cost of any damages or harm that the robot might create. 0:24:38.720 --> 0:24:42.280 So you're not paying the robot and putting money into 0:24:42.440 --> 0:24:44.760 its bank account that then goes to spend on motor 0:24:44.800 --> 0:24:48.600 oil or something. You're accumulating money in the event that 0:24:48.640 --> 0:24:52.600 the robot causes damage, and then you've got money dedicated 0:24:52.640 --> 0:24:55.320 to that robot that you can use to pay out 0:24:55.840 --> 0:25:00.400 and the event of damages being created by that robot. Uh, 0:25:00.640 --> 0:25:02.800 so you're essentially paying a robot so that in case 0:25:02.840 --> 0:25:06.399 it goes haywire and starts, I don't know, slapping people around, 0:25:06.720 --> 0:25:10.240 you can actually cover all those damages. The report also 0:25:10.280 --> 0:25:14.600 calls for specific legal status to apply to robots, thus 0:25:14.640 --> 0:25:19.679 creating assas that's equivalent to like electronic persons that have 0:25:19.800 --> 0:25:23.080 specific rights and obligations, including that of making good any 0:25:23.119 --> 0:25:26.440 damage they may cause. And a lot of people say 0:25:26.520 --> 0:25:29.359 that this is confusing, that why would you grant this 0:25:29.440 --> 0:25:33.159 personhood to robots? And a frequent response to that, and 0:25:33.200 --> 0:25:35.280 I think it's got some validity to it as well. 0:25:35.320 --> 0:25:38.440 We already do it with businesses. We all are already 0:25:38.480 --> 0:25:43.640 grant the concept of personhood to corporations. Corporations can behave 0:25:43.720 --> 0:25:48.119 in legal matters as if they are people, human being people. 0:25:48.680 --> 0:25:51.359 And if a corporation can do it a business a 0:25:51.560 --> 0:25:55.880 collection of work, then why not a robot. Then there's 0:25:55.880 --> 0:25:58.399 a section from the design of an Ethical Framework to 0:25:58.440 --> 0:26:01.840 make certain advances that uh that are made in robotics 0:26:01.960 --> 0:26:08.320 are made with consideration to how it impacts human safety, privacy, dignity, 0:26:08.840 --> 0:26:12.600 your who owns information. This proposal goes on to say 0:26:12.640 --> 0:26:15.280 that the risk of harm should be no greater than 0:26:15.480 --> 0:26:18.919 encountered in ordinary life. So, in other words, a future 0:26:18.920 --> 0:26:21.359