WEBVTT 00:00:14.640 --> 00:00:17.710 align:middle line:84% I just want to reassure you this isn't about blame. 00:00:17.710 --> 00:00:20.470 align:middle line:84% It's about using what happened to all of us, 00:00:20.470 --> 00:00:22.360 align:middle line:84% this thing we all just lived through 00:00:22.360 --> 00:00:25.180 align:middle line:84% - those of us who did live through it. 00:00:25.180 --> 00:00:29.605 align:middle line:84% It's about using it as a teaching moment. 00:00:29.605 --> 00:00:31.970 align:middle line:90% I've always hated that phrase. 00:00:31.970 --> 00:00:32.470 align:middle line:90% Me too. 00:00:42.520 --> 00:00:44.133 align:middle line:90% Basically, you're not on trial. 00:00:44.133 --> 00:00:46.300 align:middle line:84% Your company is one of many nationwide who signed up 00:00:46.300 --> 00:00:49.240 align:middle line:84% to this project to map basic human behaviour during 00:00:49.240 --> 00:00:52.990 align:middle line:84% the pandemic, both for academia, research, 00:00:52.990 --> 00:00:56.510 align:middle line:84% also to help us plan for future crises using all the... 00:00:56.510 --> 00:00:57.010 align:middle line:90% Sorry. 00:00:57.010 --> 00:00:58.570 align:middle line:90% Us being? 00:00:58.570 --> 00:01:00.880 align:middle line:84% Oh, just your average software enterprise company, 00:01:00.880 --> 00:01:03.550 align:middle line:84% one of the many private firms contracted during the crisis 00:01:03.550 --> 00:01:06.490 align:middle line:84% to help the government in gathering data 00:01:06.490 --> 00:01:08.095 align:middle line:90% to inform policy. 00:01:08.095 --> 00:01:10.300 align:middle line:84% To find the worst lockdown offenders, you mean? 00:01:10.300 --> 00:01:14.515 align:middle line:84% You're trying to pinpoint who is most responsible. 00:01:14.515 --> 00:01:16.375 align:middle line:84% That's one way of looking at it, yeah. 00:01:16.375 --> 00:01:18.520 align:middle line:84% What's the other way of looking at it? 00:01:18.520 --> 00:01:20.921 align:middle line:90% Who was most irresponsible. 00:01:23.810 --> 00:01:25.465 align:middle line:90% How was it for you? 00:01:25.465 --> 00:01:27.790 align:middle line:90% How was my existential dystopia? 00:01:27.790 --> 00:01:28.442 align:middle line:90% Yeah. 00:01:28.442 --> 00:01:29.970 align:middle line:90% It was all right. 00:01:29.970 --> 00:01:32.290 align:middle line:90% How was yours? 00:01:32.290 --> 00:01:32.790 align:middle line:90% Yeah. 00:01:32.790 --> 00:01:35.350 align:middle line:84% Well, my fiancé decided to ride that last one out with his sick 00:01:35.350 --> 00:01:35.850 align:middle line:90% mum. 00:01:35.850 --> 00:01:41.633 align:middle line:84% So not amazing, amazing, but fine. 00:01:41.633 --> 00:01:43.300 align:middle line:84% The third British lockdown came to force 00:01:43.300 --> 00:01:45.670 align:middle line:90% on 6th of January 2021. 00:01:45.670 --> 00:01:47.890 align:middle line:90% You began working from home. 00:01:47.890 --> 00:01:51.600 align:middle line:90% You're a barrister. 00:01:51.600 --> 00:01:54.810 align:middle line:84% Junior, four years call, family law. 00:01:54.810 --> 00:01:56.390 align:middle line:90% Cool. 00:01:56.390 --> 00:01:59.000 align:middle line:84% Well, like a lot of firms yours installed software 00:01:59.000 --> 00:02:01.820 align:middle line:84% to monitor when people were away from their desks. 00:02:01.820 --> 00:02:04.370 align:middle line:84% You were pretty consistent at first, 00:02:04.370 --> 00:02:11.185 align:middle line:84% but then your patterns started to change. 00:02:11.185 --> 00:02:11.927 align:middle line:90% Did they? 00:02:11.927 --> 00:02:12.470 align:middle line:90% Yeah. 00:02:12.470 --> 00:02:12.970 align:middle line:90% Yeah. 00:02:12.970 --> 00:02:14.860 align:middle line:84% That first week, your smartwatch recorded 00:02:14.860 --> 00:02:18.685 align:middle line:84% you going on a nice run at 7:00 every morning. 00:02:18.685 --> 00:02:19.502 align:middle line:90% You have that? 00:02:19.502 --> 00:02:22.810 align:middle line:84% 4K, which took you an average of 21 minutes. 00:02:22.810 --> 00:02:23.980 align:middle line:90% Not bad. 00:02:23.980 --> 00:02:26.440 align:middle line:84% But then, from January the 18th onwards 00:02:26.440 --> 00:02:28.810 align:middle line:84% you stopped that routine, and you 00:02:28.810 --> 00:02:33.240 align:middle line:84% started to leave the house at night 00:02:33.240 --> 00:02:36.530 align:middle line:90% instead, which is still OK. 00:02:36.530 --> 00:02:37.475 align:middle line:90% It's just... 00:02:37.475 --> 00:02:39.050 align:middle line:90% Is it OK? 00:02:39.050 --> 00:02:41.840 align:middle line:84% Because it feels like I might be being made to feel like it's 00:02:41.840 --> 00:02:43.595 align:middle line:90% not OK. 00:02:43.595 --> 00:02:45.657 align:middle line:84% Like I said, this is not about judgement. 00:02:45.657 --> 00:02:46.490 align:middle line:90% It's about patterns. 00:02:49.100 --> 00:02:51.500 align:middle line:84% You live with your fiancé, Jason, 00:02:51.500 --> 00:02:54.095 align:middle line:84% and you rent a spare room to Alison. 00:02:54.095 --> 00:02:56.720 align:middle line:84% You've got my housemate's name from my smartwatch? 00:02:56.720 --> 00:03:00.120 align:middle line:84% Your cohabitants' names are in the local authority's database. 00:03:00.120 --> 00:03:04.190 align:middle line:84% Our firm has partnered up with many different bodies 00:03:04.190 --> 00:03:10.730 align:middle line:84% as part of our work to collate both public and private 00:03:10.730 --> 00:03:12.060 align:middle line:90% consumer data. 00:03:12.060 --> 00:03:12.560 align:middle line:90% Yeah. 00:03:12.560 --> 00:03:15.143 align:middle line:84% It's just that this is starting to feel like an interrogation. 00:03:15.143 --> 00:03:18.035 align:middle line:90% Only, you're not the police. 00:03:18.035 --> 00:03:19.847 align:middle line:90% It's not an interrogation. 00:03:19.847 --> 00:03:21.680 align:middle line:84% Although, we have partnered with the police. 00:03:21.680 --> 00:03:22.835 align:middle line:90% And actually... 00:03:25.400 --> 00:03:29.630 align:middle line:84% this isn't me being... but the police now come to software 00:03:29.630 --> 00:03:34.820 align:middle line:84% companies like us for intel, not the other way around. 00:03:34.820 --> 00:03:37.100 align:middle line:84% To the evening of January the 19th, then, 00:03:37.100 --> 00:03:40.730 align:middle line:84% you left the house in non-running gear without 00:03:40.730 --> 00:03:48.941 align:middle line:84% your smartwatch or your smartphone, almost like... 00:03:48.941 --> 00:03:52.950 align:middle line:84% I don't know... like you didn't want to be tracked. 00:03:52.950 --> 00:03:54.630 align:middle line:90% Yet you were tracking me. 00:03:54.630 --> 00:03:56.640 align:middle line:90% No one was tracking you, then. 00:03:56.640 --> 00:03:58.488 align:middle line:84% This is retrospective, historic data. 00:03:58.488 --> 00:04:00.405 align:middle line:84% How'd you know I wasn't wearing my smartwatch? 00:04:00.405 --> 00:04:01.190 align:middle line:90% A door cam. 00:04:01.190 --> 00:04:02.385 align:middle line:90% I don't have a door cam. 00:04:02.385 --> 00:04:04.710 align:middle line:84% You don't have a door cam, but your neighbour has one. 00:04:04.710 --> 00:04:07.395 align:middle line:90% A Ring door cam, it caught you. 00:04:07.395 --> 00:04:11.685 align:middle line:84% Ring just gave you access to any... 00:04:11.685 --> 00:04:13.252 align:middle line:90% The police partnered with Ring. 00:04:13.252 --> 00:04:14.460 align:middle line:90% We partnered with the police. 00:04:14.460 --> 00:04:16.793 align:middle line:84% Look, the point is, Rachel, can I ask you where you went 00:04:16.793 --> 00:04:18.120 align:middle line:90% and why? 00:04:18.120 --> 00:04:20.279 align:middle line:90% Just for this study, that's all. 00:04:20.279 --> 00:04:23.222 align:middle line:84% We didn't pick you up via any facial recognition on CCTV, 00:04:23.222 --> 00:04:24.930 align:middle line:84% probably because you were wearing a mask. 00:04:24.930 --> 00:04:25.430 align:middle line:90% Very good. 00:04:25.430 --> 00:04:27.090 align:middle line:84% Although, I'm not going to lie, that 00:04:27.090 --> 00:04:29.970 align:middle line:84% has been a bit of a nightmare in my field, let me tell you. 00:04:29.970 --> 00:04:33.990 align:middle line:84% I have no idea where... who the hell remembers this stuff? 00:04:33.990 --> 00:04:39.840 align:middle line:84% You repeated these little jaunts for about a week, sometimes 00:04:39.840 --> 00:04:41.110 align:middle line:90% with your smartwatch on now. 00:04:41.110 --> 00:04:44.400 align:middle line:84% Maybe you forgot to take it off, easily done. 00:04:44.400 --> 00:04:50.140 align:middle line:84% You went to roughly this location. 00:04:50.140 --> 00:04:52.960 align:middle line:90% What was there? 00:04:52.960 --> 00:04:53.650 align:middle line:90% Who was there? 00:04:57.781 --> 00:05:00.810 align:middle line:90% Do you know what I think? 00:05:00.810 --> 00:05:05.032 align:middle line:90% I think you missed your fiancé. 00:05:05.032 --> 00:05:05.640 align:middle line:90% Do you? 00:05:05.640 --> 00:05:09.630 align:middle line:84% I think this was an extraordinary, traumatic time, 00:05:09.630 --> 00:05:11.953 align:middle line:90% and it was hard. 00:05:11.953 --> 00:05:13.620 align:middle line:84% The real world was shuttered up, leaving 00:05:13.620 --> 00:05:21.810 align:middle line:84% only the digital world and its offer of new temptations. 00:05:21.810 --> 00:05:24.670 align:middle line:84% I'm not a psychologist or a sociologist. 00:05:24.670 --> 00:05:25.540 align:middle line:90% I'm an analyst. 00:05:25.540 --> 00:05:29.880 align:middle line:84% But, judging by the data and, frankly, 00:05:29.880 --> 00:05:34.105 align:middle line:84% my own personal observations, I think 00:05:34.105 --> 00:05:36.480 align:middle line:84% the boundaries were crossed by many of us, by most of us, 00:05:36.480 --> 00:05:40.920 align:middle line:84% understandably, even the prudish Brits with messages that 00:05:40.920 --> 00:05:45.840 align:middle line:84% become images that become videos that become 00:05:45.840 --> 00:05:57.890 align:middle line:84% live video, our hidden fantasies and exposed faces, even 00:05:57.890 --> 00:06:00.420 align:middle line:84% though we vaguely know deep down that these images we send, 00:06:00.420 --> 00:06:02.937 align:middle line:84% they're not going straight to the other person's device, 00:06:02.937 --> 00:06:04.520 align:middle line:84% even though that's what it feels like, 00:06:04.520 --> 00:06:06.390 align:middle line:90% intimate one-to-one connection. 00:06:06.390 --> 00:06:06.890 align:middle line:90% No. 00:06:06.890 --> 00:06:10.160 align:middle line:84% They're instead being handed over to a tech company 00:06:10.160 --> 00:06:12.490 align:middle line:84% to store on a server that's probably 00:06:12.490 --> 00:06:14.240 align:middle line:84% in a different country with different laws 00:06:14.240 --> 00:06:16.400 align:middle line:84% where it will sit possibly forever. 00:06:16.400 --> 00:06:17.060 align:middle line:90% We know that. 00:06:17.060 --> 00:06:21.922 align:middle line:90% But, we're also only human. 00:06:21.922 --> 00:06:23.880 align:middle line:84% I think it's also because the home changed too. 00:06:23.880 --> 00:06:24.240 align:middle line:90% Didn't it? 00:06:24.240 --> 00:06:26.040 align:middle line:84% Literally, the nature of what the home had 00:06:26.040 --> 00:06:30.540 align:middle line:84% been for centuries, gone within weeks, from a private place, 00:06:30.540 --> 00:06:35.600 align:middle line:84% a sanctuary, to a public space of work. 00:06:35.600 --> 00:06:40.190 align:middle line:84% And we were forced to welcome anyone and everyone into it 00:06:40.190 --> 00:06:41.285 align:middle line:90% through a screen. 00:06:41.285 --> 00:06:44.630 align:middle line:84% But I didn't send anything to Jason or... 00:06:44.630 --> 00:06:46.020 align:middle line:90% Or? 00:06:46.020 --> 00:06:48.090 align:middle line:84% Are you really sitting here suggesting I betrayed 00:06:48.090 --> 00:06:50.325 align:middle line:90% my fiancé? 00:06:50.325 --> 00:06:54.150 align:middle line:84% We know you started to develop huge feelings of anxiety 00:06:54.150 --> 00:06:57.330 align:middle line:90% and guilt about something. 00:06:57.330 --> 00:06:58.470 align:middle line:90% OK? 00:06:58.470 --> 00:07:01.620 align:middle line:84% You can read all the location settings and datasets you like. 00:07:01.620 --> 00:07:05.490 align:middle line:84% You can't get inside my head, my heart. 00:07:05.490 --> 00:07:09.510 align:middle line:84% Actually, when you returned home from that first evening 00:07:09.510 --> 00:07:12.600 align:middle line:84% away from the house, your smart home thermostat 00:07:12.600 --> 00:07:19.170 align:middle line:84% registered you showering for a full 20 minutes. 00:07:19.170 --> 00:07:21.430 align:middle line:90% Trying to wash something away? 00:07:21.430 --> 00:07:22.195 align:middle line:90% Excuse me. 00:07:22.195 --> 00:07:25.000 align:middle line:84% That night and many nights following, you 00:07:25.000 --> 00:07:28.540 align:middle line:90% had a deeply restless sleep. 00:07:28.540 --> 00:07:30.880 align:middle line:84% You asked Alexa to list Covid symptoms on the morning 00:07:30.880 --> 00:07:32.920 align:middle line:90% of the 21st at 9:34. 00:07:32.920 --> 00:07:35.538 align:middle line:84% And Alexa returned results to you from the NHS website. 00:07:35.538 --> 00:07:36.955 align:middle line:84% That could have been my housemate. 00:07:36.955 --> 00:07:37.600 align:middle line:90% It was your voice. 00:07:37.600 --> 00:07:40.142 align:middle line:84% And by the way, your Alexa, your Echo, your Siri, your Google 00:07:40.142 --> 00:07:42.040 align:middle line:90% Dot, they are always listening. 00:07:42.040 --> 00:07:45.040 align:middle line:84% Why, I wonder, did you have to check symptoms 00:07:45.040 --> 00:07:48.070 align:middle line:84% if you hadn't been close to anyone outside your house? 00:07:48.070 --> 00:07:49.900 align:middle line:84% And why were you googling questions 00:07:49.900 --> 00:07:52.990 align:middle line:84% around forgiveness, Rachel, at 3:00 am in the morning? 00:07:52.990 --> 00:07:54.370 align:middle line:90% How to forgive? 00:07:54.370 --> 00:07:58.150 align:middle line:84% Can you forgive someone who hurt you? 00:07:58.150 --> 00:07:58.840 align:middle line:90% Jesus Christ! 00:07:58.840 --> 00:08:00.113 align:middle line:90% This is some Stasi level... 00:08:00.113 --> 00:08:01.780 align:middle line:84% And then, of course, there's the contact 00:08:01.780 --> 00:08:04.360 align:middle line:84% tracing app, which all dutiful citizens downloaded. 00:08:04.360 --> 00:08:08.410 align:middle line:84% Well, that tells us that you were in the presence of someone 00:08:08.410 --> 00:08:11.350 align:middle line:84% who also had the app for sustained periods, 00:08:11.350 --> 00:08:17.740 align:middle line:84% a man who wasn't your fiancé, a man who when you were in his 00:08:17.740 --> 00:08:20.200 align:middle line:84% company that first night that you were in his physical 00:08:20.200 --> 00:08:31.740 align:middle line:84% proximity, your heartbeat went up, and up, and up. 00:08:31.740 --> 00:08:35.929 align:middle line:84% So you see, it is possible to see into your heart, 00:08:35.929 --> 00:08:40.560 align:middle line:84% maybe even to see what you refuse to acknowledge yourself. 00:08:40.560 --> 00:08:41.590 align:middle line:90% I no longer consent. 00:08:41.590 --> 00:08:42.950 align:middle line:90% This is beyond the remit. 00:08:42.950 --> 00:08:44.070 align:middle line:90% You have no right! 00:08:44.070 --> 00:08:44.880 align:middle line:90% Right? 00:08:44.880 --> 00:08:47.460 align:middle line:84% We are trying to understand what motivated people to break 00:08:47.460 --> 00:08:48.978 align:middle line:84% the restriction, and if it's love.. 00:08:48.978 --> 00:08:49.710 align:middle line:84% I didn't intentionally break anything. 00:08:49.710 --> 00:08:50.256 align:middle line:90% ...if it's love... 00:08:50.256 --> 00:08:50.883 align:middle line:90% If I could explain... 00:08:50.883 --> 00:08:51.720 align:middle line:90% ...or lust or fear... 00:08:51.720 --> 00:08:52.900 align:middle line:84% Actually, no, I shouldn't have to explain. 00:08:52.900 --> 00:08:55.020 align:middle line:84% ...or anything we have to plan to mitigate these human flaws 00:08:55.020 --> 00:08:57.600 align:middle line:84% coded into us, feelings that make us act against our own 00:08:57.600 --> 00:08:59.190 align:middle line:84% best interest and the interest... 00:08:59.190 --> 00:08:59.910 align:middle line:84% I don't really think an anonymous... 00:08:59.910 --> 00:09:00.420 align:middle line:90% ...of those around us. 00:09:00.420 --> 00:09:01.005 align:middle line:90% ...data storage company... 00:09:01.005 --> 00:09:01.360 align:middle line:90% Was it us? 00:09:01.360 --> 00:09:03.240 align:middle line:84% Is there something in our national character that.. 00:09:03.240 --> 00:09:04.470 align:middle line:84% .funded by God knows poking around in people's have any 00:09:04.470 --> 00:09:05.695 align:middle line:90% right to moralise over... 00:09:05.695 --> 00:09:06.930 align:middle line:84% ...meant that we topped the list of highest infection rates 00:09:06.930 --> 00:09:07.980 align:middle line:90% in the world? 00:09:07.980 --> 00:09:09.480 align:middle line:90% Why us? 00:09:09.480 --> 00:09:12.390 align:middle line:84% Why in the third wave was there a spike 00:09:12.390 --> 00:09:14.040 align:middle line:90% of infections in your area? 00:09:14.040 --> 00:09:14.880 align:middle line:90% Who was that? 00:09:14.880 --> 00:09:16.800 align:middle line:84% Who was the patient zero of this story? 00:09:16.800 --> 00:09:18.700 align:middle line:84% Whose behaviour started this spread? 00:09:18.700 --> 00:09:19.200 align:middle line:90% Hmm? 00:09:19.200 --> 00:09:21.195 align:middle line:84% Have you ever seen those maps in school books of... 00:09:21.195 --> 00:09:21.382 align:middle line:90% You're blaming... 00:09:21.382 --> 00:09:22.957 align:middle line:84% ..of how the Great Fire of London broke out... 00:09:22.957 --> 00:09:23.570 align:middle line:90% ...normal people. 00:09:23.570 --> 00:09:24.750 align:middle line:90% ...and spread from shop to shop? 00:09:24.750 --> 00:09:25.335 align:middle line:90% Everyone tried... 00:09:25.335 --> 00:09:26.040 align:middle line:90% Future schoolchildren... 00:09:26.040 --> 00:09:26.445 align:middle line:84% ...ordinary, under extraordinary... 00:09:26.445 --> 00:09:28.230 align:middle line:84% ..studying the great pandemic of 2020 are going to be able 00:09:28.230 --> 00:09:29.533 align:middle line:90% to draw their own map... 00:09:29.533 --> 00:09:29.686 align:middle line:90% OK. 00:09:29.686 --> 00:09:30.768 align:middle line:90% You're blaming normal people... 00:09:30.768 --> 00:09:31.450 align:middle line:90% ...of person-to-person contact. 00:09:31.450 --> 00:09:31.900 align:middle line:90% ...not the government response. 00:09:31.900 --> 00:09:32.370 align:middle line:90% We have it all. 00:09:32.370 --> 00:09:32.940 align:middle line:90% It's all stored... 00:09:32.940 --> 00:09:33.663 align:middle line:84% ...not the conspiracy theories... 00:09:33.663 --> 00:09:34.191 align:middle line:90% ..every human interaction... 00:09:34.191 --> 00:09:35.125 align:middle line:84% ...not the spread of misinformation... 00:09:35.125 --> 00:09:36.471 align:middle line:84% ...every footstep, every message... 00:09:36.471 --> 00:09:37.410 align:middle line:90% ...by your algorithm. 00:09:37.410 --> 00:09:38.610 align:middle line:90% ...every secret thought. 00:09:38.610 --> 00:09:39.480 align:middle line:90% They're going to draw a map. 00:09:39.480 --> 00:09:39.900 align:middle line:90% You're blaming people. 00:09:39.900 --> 00:09:41.215 align:middle line:84% And who, I wonder, will be at the centre of it? 00:09:41.215 --> 00:09:41.640 align:middle line:90% You're blaming people.. 00:09:41.640 --> 00:09:43.505 align:middle line:84% Who's the Typhoid Mary of this story... 00:09:43.505 --> 00:09:44.460 align:middle line:84% ...not the spread of misinformation 00:09:44.460 --> 00:09:44.996 align:middle line:90% by your algorithm. 00:09:44.996 --> 00:09:46.095 align:middle line:84% ...responsible for all those extra deaths? 00:09:46.095 --> 00:09:46.620 align:middle line:90% You're blaming people! 00:09:46.620 --> 00:09:48.537 align:middle line:84% ...responsible for the pain of losing someone, 00:09:48.537 --> 00:09:53.290 align:middle line:84% of not being able to visit your father in hospital but to say 00:09:53.290 --> 00:09:55.270 align:middle line:84% goodbye to him over a phone instead, 00:09:55.270 --> 00:09:58.870 align:middle line:84% as you hear his last breaths ebbing away over a fucking 00:09:58.870 --> 00:09:59.975 align:middle line:90% speakerphone? 00:10:04.872 --> 00:10:06.590 align:middle line:90% Apple. 00:10:06.590 --> 00:10:07.940 align:middle line:90% Reminder to stand and breathe. 00:10:44.900 --> 00:10:50.270 align:middle line:84% So I know I specialise in family law, not whatever this is. 00:10:50.270 --> 00:10:51.030 align:middle line:90% I think... 00:10:51.030 --> 00:10:51.530 align:middle line:90% Oh, please. 00:10:51.530 --> 00:10:52.242 align:middle line:90% What are you... 00:10:52.242 --> 00:10:55.640 align:middle line:84% I think citizens have the right to express some unease 00:10:55.640 --> 00:10:59.270 align:middle line:84% at the way during a pandemic technology companies rode 00:10:59.270 --> 00:11:01.160 align:middle line:84% in to help governments all over the world, 00:11:01.160 --> 00:11:05.090 align:middle line:84% often free of charge, to collate and gain access to citizen 00:11:05.090 --> 00:11:05.870 align:middle line:90% data... 00:11:05.870 --> 00:11:07.077 align:middle line:90% To beat a virus. 00:11:07.077 --> 00:11:09.890 align:middle line:84% ...with little scrutiny or oversight. 00:11:09.890 --> 00:11:12.350 align:middle line:84% Where was the vote, the public consultation 00:11:12.350 --> 00:11:16.100 align:middle line:84% in letting Amazon pair with the NHS, a private company, 00:11:16.100 --> 00:11:17.450 align:middle line:90% our health data? 00:11:17.450 --> 00:11:20.030 align:middle line:84% But that's the point... isn't it? ...about algorithms. 00:11:20.030 --> 00:11:22.460 align:middle line:84% No one really knows how they work, 00:11:22.460 --> 00:11:24.950 align:middle line:84% even though in an increasingly automated world 00:11:24.950 --> 00:11:27.087 align:middle line:84% AI is making decisions for us all the time 00:11:27.087 --> 00:11:30.230 align:middle line:84% - what our insurance premiums should be, 00:11:30.230 --> 00:11:32.870 align:middle line:84% or A level results, or predictive policing, 00:11:32.870 --> 00:11:35.075 align:middle line:84% or even prison sentencing and release. 00:11:35.075 --> 00:11:37.130 align:middle line:84% All of it auditable and accountable. 00:11:37.130 --> 00:11:38.570 align:middle line:90% Is it accountable? 00:11:38.570 --> 00:11:40.608 align:middle line:84% What if, say, I had a friend that 00:11:40.608 --> 00:11:42.650 align:middle line:84% had a daughter that lived in a residential street 00:11:42.650 --> 00:11:44.900 align:middle line:84% that one day, a Maps application discovered 00:11:44.900 --> 00:11:46.820 align:middle line:84% was quiet and redirected cars down it 00:11:46.820 --> 00:11:49.520 align:middle line:84% to avoid a build up of traffic elsewhere? 00:11:49.520 --> 00:11:52.108 align:middle line:84% And my friend's daughter, she runs out into the road, 00:11:52.108 --> 00:11:53.400 align:middle line:90% because it was quiet yesterday. 00:11:53.400 --> 00:11:55.610 align:middle line:84% So she thinks it's going to be quiet today. 00:11:55.610 --> 00:11:59.453 align:middle line:84% And she's knocked down, and she's killed. 00:11:59.453 --> 00:12:01.620 align:middle line:84% And my friend wants justice but it is denied to her, 00:12:01.620 --> 00:12:05.780 align:middle line:84% because you cannot take an algorithm to court. 00:12:05.780 --> 00:12:08.840 align:middle line:84% OK, what we do know about these so-called digital gods 00:12:08.840 --> 00:12:12.530 align:middle line:84% playing with our lives is that unlike what you said earlier, 00:12:12.530 --> 00:12:15.470 align:middle line:84% they do contain all the human flaws of the people who coded 00:12:15.470 --> 00:12:16.220 align:middle line:90% and created them. 00:12:16.220 --> 00:12:18.930 align:middle line:84% Was it Katy O'Neil, Cathy O'Neil, 00:12:18.930 --> 00:12:22.640 align:middle line:84% who said, algorithms are simply opinions embedded in code? 00:12:22.640 --> 00:12:25.610 align:middle line:84% Facial technology didn't pick up my face because of my mask 00:12:25.610 --> 00:12:27.680 align:middle line:84% or because it's up to 40 per cent less effective 00:12:27.680 --> 00:12:30.530 align:middle line:84% at recognising black women than white men? 00:12:30.530 --> 00:12:32.750 align:middle line:84% And you're asking me to submit myself 00:12:32.750 --> 00:12:34.492 align:middle line:84% to the infallibility of a system that 00:12:34.492 --> 00:12:36.200 align:middle line:84% doesn't allow anyone to pop open the hood 00:12:36.200 --> 00:12:37.380 align:middle line:90% and take a look inside. 00:12:37.380 --> 00:12:39.590 align:middle line:84% I do not submit myself to it, to any of it. 00:12:39.590 --> 00:12:41.970 align:middle line:84% Not everything can be known from data. 00:12:41.970 --> 00:12:43.110 align:middle line:90% People are individual. 00:12:43.110 --> 00:12:43.925 align:middle line:90% We are complex. 00:12:51.015 --> 00:12:54.400 align:middle line:84% You blame the government in outsourcing its decisions 00:12:54.400 --> 00:12:57.040 align:middle line:90% to data companies like ours. 00:12:57.040 --> 00:12:59.830 align:middle line:84% You outsource every day of your life. 00:12:59.830 --> 00:13:04.756 align:middle line:84% You like this or try that, where to eat, what to watch. 00:13:04.756 --> 00:13:07.960 align:middle line:84% You found your fiancé on a dating app. 00:13:07.960 --> 00:13:09.940 align:middle line:90% You outsourced your love. 00:13:09.940 --> 00:13:11.140 align:middle line:90% Oh, pick me, someone! 00:13:11.140 --> 00:13:11.830 align:middle line:90% Pick me! 00:13:11.830 --> 00:13:13.840 align:middle line:84% You handed over your own human agency. 00:13:13.840 --> 00:13:16.450 align:middle line:84% So nice speech, but most people very willingly 00:13:16.450 --> 00:13:19.810 align:middle line:84% during that pandemic sacrificed a few very outdated notions 00:13:19.810 --> 00:13:21.460 align:middle line:84% of privacy, so that the algorithms 00:13:21.460 --> 00:13:26.355 align:middle line:84% could do what people couldn't, track the virus and save lives. 00:13:26.355 --> 00:13:27.730 align:middle line:84% And I know that you like to think 00:13:27.730 --> 00:13:28.870 align:middle line:90% that everyone's an individual. 00:13:28.870 --> 00:13:29.495 align:middle line:90% But you're not. 00:13:29.495 --> 00:13:30.640 align:middle line:90% I'm not. 00:13:30.640 --> 00:13:34.090 align:middle line:84% I fall into a predictable pattern of behaviours based 00:13:34.090 --> 00:13:36.880 align:middle line:84% on 100 different data points based on my race, my age, 00:13:36.880 --> 00:13:38.973 align:middle line:90% my class, the things I watch. 00:13:38.973 --> 00:13:40.390 align:middle line:84% You think you're one in a million? 00:13:40.390 --> 00:13:41.860 align:middle line:90% Congratulations! 00:13:41.860 --> 00:13:43.840 align:middle line:84% On a planet of 8 billion, that means 00:13:43.840 --> 00:13:47.680 align:middle line:84% there are more than 8,000 people exactly like you. 00:13:47.680 --> 00:13:50.590 align:middle line:84% Did you break the restrictions because of a relationship 00:13:50.590 --> 00:13:51.235 align:middle line:90% with this man? 00:13:51.235 --> 00:13:52.818 align:middle line:84% All you can measure is what people do! 00:13:52.818 --> 00:13:53.845 align:middle line:90% You can't know why! 00:13:53.845 --> 00:13:55.270 align:middle line:90% There is no why good enough. 00:13:55.270 --> 00:13:56.080 align:middle line:90% Did you? 00:13:56.080 --> 00:13:59.020 align:middle line:84% Did you have an affair with Joe, a man whose Facebook profile 00:13:59.020 --> 00:14:01.630 align:middle line:84% you clicked on 28 times in 24 hours, 00:14:01.630 --> 00:14:03.876 align:middle line:84% a man who went home to contract Covid himself... 00:14:03.876 --> 00:14:04.100 align:middle line:90% It's true. 00:14:04.100 --> 00:14:04.746 align:middle line:90% I don't trust this... 00:14:04.746 --> 00:14:05.753 align:middle line:84% ...and then spread it to at least 12 others... 00:14:05.753 --> 00:14:07.240 align:middle line:84% ...but not because your algorithms aren't good 00:14:07.240 --> 00:14:07.666 align:middle line:90% enough... 00:14:07.666 --> 00:14:08.985 align:middle line:84% ...who spread it to at least 58 others? 00:14:08.985 --> 00:14:09.340 align:middle line:90% ...but because they're too good. 00:14:09.340 --> 00:14:09.850 align:middle line:90% I mean, imagine. 00:14:09.850 --> 00:14:10.360 align:middle line:90% Imagine. 00:14:10.360 --> 00:14:12.550 align:middle line:84% Can you imagine a social media platform 00:14:12.550 --> 00:14:14.380 align:middle line:84% so good at recommending to you people 00:14:14.380 --> 00:14:17.110 align:middle line:84% that you may know that it recommends you, out of nowhere 00:14:17.110 --> 00:14:20.020 align:middle line:84% when you're nowhere near ready, the man who sexually 00:14:20.020 --> 00:14:26.350 align:middle line:84% and emotionally abused you over many months and years, who 00:14:26.350 --> 00:14:29.848 align:middle line:84% during your relationship gaslit, and tormented, and manipulated, 00:14:29.848 --> 00:14:32.140 align:middle line:84% and yet, you somehow managed to extricate yourself from 00:14:32.140 --> 00:14:35.170 align:middle line:84% to survive only to be faced with the face of that person 00:14:35.170 --> 00:14:36.520 align:middle line:90% you've been trying to forget? 00:14:36.520 --> 00:14:37.510 align:middle line:90% People you may know? 00:14:37.510 --> 00:14:40.284 align:middle line:90% I'll fucking say. 00:14:40.284 --> 00:14:42.282 align:middle line:90% And where was my control? 00:14:42.282 --> 00:14:43.115 align:middle line:90% Where was my agency? 00:14:49.074 --> 00:14:52.810 align:middle line:90% That's who you went to see? 00:14:52.810 --> 00:14:53.530 align:middle line:90% Why? 00:14:53.530 --> 00:14:54.452 align:middle line:90% To confront him? 00:14:54.452 --> 00:14:56.200 align:middle line:84% You think you'll know how you'll react. 00:14:56.200 --> 00:14:58.195 align:middle line:90% I still, under the guidelines... 00:14:58.195 --> 00:15:00.070 align:middle line:84% Based on previous behaviours, you did this. 00:15:00.070 --> 00:15:01.270 align:middle line:90% So you'll likely do that. 00:15:01.270 --> 00:15:02.187 align:middle line:90% I still don't think... 00:15:02.187 --> 00:15:06.280 align:middle line:84% You think you'll want some sort of justice, revenge. 00:15:06.280 --> 00:15:09.880 align:middle line:84% But when you enter their world, when you read what they're 00:15:09.880 --> 00:15:14.170 align:middle line:84% writing, see them struggling, like we all were, 00:15:14.170 --> 00:15:15.670 align:middle line:84% posting thoughts about ending it all 00:15:15.670 --> 00:15:17.200 align:middle line:84% and seeing no one come to their aid, 00:15:17.200 --> 00:15:22.730 align:middle line:84% well, sometimes you can surprise even yourself. 00:15:22.730 --> 00:15:26.360 align:middle line:84% Perhaps the way to move on from something done by someone, 00:15:26.360 --> 00:15:28.610 align:middle line:84% perhaps helping to save that person 00:15:28.610 --> 00:15:31.117 align:middle line:84% might just be the most surprising way of surviving 00:15:31.117 --> 00:15:31.950 align:middle line:90% that thing yourself. 00:15:37.070 --> 00:15:37.695 align:middle line:90% How to forgive? 00:15:41.795 --> 00:15:43.226 align:middle line:90% People are unpredictable. 00:15:46.235 --> 00:15:46.735 align:middle line:90% Possibly. 00:15:49.440 --> 00:15:50.550 align:middle line:90% But viruses aren't. 00:15:57.380 --> 00:16:04.820 align:middle line:84% I'll note for the record that you made a personal decision 00:16:04.820 --> 00:16:13.670 align:middle line:84% to break the restrictions to possibly save a life. 00:16:13.670 --> 00:16:15.620 align:middle line:90% But I'm sorry. 00:16:15.620 --> 00:16:18.740 align:middle line:84% That may have killed 20 more by transmitting the virus. 00:16:18.740 --> 00:16:21.330 align:middle line:84% And that's why the data knows best. 00:16:21.330 --> 00:16:25.010 align:middle line:84% And that's why we must hand ourselves over to it. 00:16:25.010 --> 00:16:28.010 align:middle line:84% As emotional beings we are simply 00:16:28.010 --> 00:16:30.510 align:middle line:84% unable to measure that equivalence ourselves. 00:16:30.510 --> 00:16:36.550 align:middle line:84% So your audit trail will enter the system, and an automated 00:16:36.550 --> 00:16:40.540 align:middle line:84% verdict delivered to you in due course. 00:16:40.540 --> 00:16:43.078 align:middle line:90% Will I be able to appeal? 00:16:43.078 --> 00:16:44.290 align:middle line:90% Yeah, of course. 00:16:44.290 --> 00:16:47.080 align:middle line:84% Although, the outcome will be the same. 00:16:47.080 --> 00:16:50.340 align:middle line:84% Algorithms, they don't tend to change their minds. 00:16:53.263 --> 00:16:54.430 align:middle line:90% Would you like an Uber home? 00:16:58.224 --> 00:17:01.555 align:middle line:90% It's recommending I walk. 00:17:01.555 --> 00:17:05.060 align:middle line:84% Well, if that's what it recommends.