WEBVTT - AI and Misconceptions: A Conversation with Sleepwalkers 0:00:04.240 --> 0:00:07.240 Welcome to Tech Stuff, a production of I Heart Radios 0:00:07.320 --> 0:00:14.200 How Stuff Works. Hey there, and welcome to tech Stuff. 0:00:14.240 --> 0:00:17.040 I'm your host, Jonathan Strickling them an executive producer with 0:00:17.079 --> 0:00:18.919 How Stuff Works in I Heart Radio and I Love 0:00:19.000 --> 0:00:21.960 all Things Tech. And you know, guys, there's no shortage 0:00:22.040 --> 0:00:26.040 of scenarios in which AI proves to be our downfall. 0:00:26.560 --> 0:00:29.600 You've got popular films like The Terminator and Matrix series 0:00:29.640 --> 0:00:34.040 in which we have artificial intelligence literally revolting against us 0:00:34.080 --> 0:00:39.640 and then subjugating us. To the numerous predictions that automation 0:00:39.720 --> 0:00:41.879 is going to displace every job, and we run the 0:00:41.920 --> 0:00:46.000 gamut of all these different scenarios where AI is going 0:00:46.040 --> 0:00:50.280 to be our end And then we have various companies 0:00:50.280 --> 0:00:53.600 and organizations that are investing billions of dollars to develop 0:00:53.800 --> 0:00:56.720 an advance artificial intelligence who are saying, no, no, no, 0:00:56.720 --> 0:00:58.080 no, no no no, you don't need to worry about that. 0:00:58.120 --> 0:01:01.040 AI is not gonna totally story of the world. It's 0:01:01.040 --> 0:01:03.160 gonna make our world better. It's going to take over 0:01:03.200 --> 0:01:05.680 the more repetitive, dull, and dangerous parts of our jobs, 0:01:05.680 --> 0:01:07.480 and it's going to free us up to concentrate on 0:01:07.520 --> 0:01:12.760 more rewarding activities. So can we get to any truth 0:01:12.840 --> 0:01:15.280 in the matter? Is there some sort of truth we 0:01:15.319 --> 0:01:19.479 can suss out from these extremes? Well, today I'm joined 0:01:19.520 --> 0:01:22.840 by Oz and Kara, the hosts of the series Sleepwalkers, 0:01:23.040 --> 0:01:25.920 a show all about AI and if you haven't checked 0:01:25.959 --> 0:01:28.880 it out yet, I highly recommend you do because it 0:01:29.000 --> 0:01:33.000 is a phenomenal show. Guys, welcome to tech Stuff. Hi, 0:01:33.720 --> 0:01:36.280 thank you for having us. Yeah, thank you so much. Jonathan. 0:01:36.280 --> 0:01:39.399 We're huge fans of tech stuff and delights to be 0:01:39.480 --> 0:01:42.280 joining the house stuff works. My heart family in the 0:01:42.400 --> 0:01:44.600 and and made part of the tech Stuff networks. So 0:01:44.720 --> 0:01:48.200 thank you. Well, thank you because you know you've you 0:01:48.280 --> 0:01:51.880 have lifted up the boat of tech stuff, certainly because 0:01:51.960 --> 0:01:56.280 your work is really inspiring. Before we jump into this conversation, 0:01:56.320 --> 0:01:58.040 if you could just take a couple of moments and 0:01:58.120 --> 0:02:01.280 let my listeners know kind of you know what Sleepwalkers is, 0:02:01.320 --> 0:02:03.520 how you would describe that show to somebody. Let's say 0:02:03.520 --> 0:02:07.080 you're at a cocktail party and you are asked what 0:02:07.120 --> 0:02:08.560 do you do for a living? You say, well, I'm 0:02:08.600 --> 0:02:11.120 working on this show. How do you describe it? So 0:02:11.160 --> 0:02:13.080 they called I think they call it an elevator pitch. 0:02:13.160 --> 0:02:16.280 But this is a cocktail pitch. Yeah, and were based 0:02:16.280 --> 0:02:17.639 in New York, so we spend a whole lives of 0:02:17.680 --> 0:02:19.800 cocktail parties. I was born at a cocktail party and 0:02:19.800 --> 0:02:22.360 committee which happened in elevators in New York, as I 0:02:22.400 --> 0:02:27.440 understand exactly, or apartments, the size of elevators. Um. So, 0:02:27.880 --> 0:02:32.880 Sleepwalkers is a podcast that actually Oz came to me with. 0:02:33.440 --> 0:02:38.200 It was his idea, his brain child. But I will 0:02:38.240 --> 0:02:40.880 say first, you know, I've I used to report on 0:02:41.120 --> 0:02:44.280 tech and science at the Huffington Post, and I had 0:02:44.280 --> 0:02:47.240 a show called Talk Nerdy to me and when Oz 0:02:47.280 --> 0:02:48.600 came to me and said, you know, I want to 0:02:48.600 --> 0:02:51.240 I want to really make a show that deals with 0:02:51.840 --> 0:02:56.400 all of the human touch points that AI could possibly 0:02:56.440 --> 0:03:02.160 come in contact with, so healthcare, agriculture or uh, you know, 0:03:02.280 --> 0:03:06.919 science in general. Love you know, all of these places 0:03:07.000 --> 0:03:10.520 where people aren't necessarily thinking that a I will have 0:03:10.560 --> 0:03:15.520 an impact, but they already should be basically, And you know, 0:03:15.560 --> 0:03:18.320 I said yes very quickly because I'm very interested in 0:03:18.320 --> 0:03:21.320 all of those touch points. So each episode really is 0:03:21.360 --> 0:03:25.919 a deep dive into one of those areas, as I said, 0:03:25.919 --> 0:03:33.519 whether it be healthcare, transhumanism, agriculture, the military, for example, um, 0:03:33.560 --> 0:03:35.839 you know, these are these are places where we're going 0:03:35.840 --> 0:03:37.760 to see the presence of AI. Were already seeing the 0:03:37.800 --> 0:03:40.160 presence of AI, and and the show really tries to 0:03:40.240 --> 0:03:45.000 explore that. Yeah, I think it's uh, it's really pretty 0:03:45.040 --> 0:03:48.520 incredible when you sit down and look at where AI 0:03:48.720 --> 0:03:52.920 has already kind of crept into our day to day experience, 0:03:53.080 --> 0:03:57.440 sometimes in ways that we wouldn't necessarily associate with AI. 0:03:57.480 --> 0:04:01.840 Like one report I read said you could argue it's 0:04:01.840 --> 0:04:05.200 a very limited application of AI, but that things like 0:04:05.320 --> 0:04:08.800 spell check and grammar check, which are now standard in 0:04:09.200 --> 0:04:13.600 apps and clients and smartphones and browsers, that that's a 0:04:14.000 --> 0:04:17.400 type of artificial intelligence that if it's doing something besides 0:04:17.480 --> 0:04:21.440 just detecting No, this sequence of letters doesn't correspond with 0:04:21.480 --> 0:04:24.960 any words in the language you are writing in. It's 0:04:25.000 --> 0:04:29.280 also perhaps looking at context, like saying, well, you use 0:04:29.360 --> 0:04:32.000 the word weather, but you use the word weather as 0:04:32.000 --> 0:04:36.560 in the types of meteor logical activity that are outside 0:04:36.600 --> 0:04:39.240 the window, as opposed to whether or not you should 0:04:39.320 --> 0:04:43.440 do something. And so you think about that and you'll realize, oh, yeah, 0:04:43.440 --> 0:04:46.440 I guess, I guess there is a lot more to 0:04:46.480 --> 0:04:48.240 it than I thought, which kind of brings me to 0:04:48.320 --> 0:04:50.440 the first point I wanted to make before we dive 0:04:50.480 --> 0:04:55.159 into the various doom and gloom scenarios of AI, which is, 0:04:55.560 --> 0:04:59.400 how do you guys define artificial intelligence? Because I found 0:04:59.520 --> 0:05:04.160 that the this concept it's so broad that often you 0:05:04.200 --> 0:05:07.360 can have two people trying to have a meaningful conversation 0:05:07.400 --> 0:05:10.640 about AI and they're not able to meet in the middle. 0:05:10.680 --> 0:05:12.720 But it's not because they don't agree with each other. 0:05:12.760 --> 0:05:16.560 It's simply because they're working from vastly different definitions of 0:05:16.600 --> 0:05:21.640 what artificial intelligence actually is. I think that's a great point, Jonathan. 0:05:22.400 --> 0:05:24.280 Just to back up a little bit, UM, I want 0:05:24.279 --> 0:05:27.120 to tell you how I count with the name Sleepwalkers 0:05:27.120 --> 0:05:29.120 for the series UM, and then I'd love to dive 0:05:29.160 --> 0:05:31.760 into I think the excellent point you make, which is 0:05:31.800 --> 0:05:38.200 that effectively yesterday's AI is today is computing. UM. But 0:05:38.520 --> 0:05:42.160 so I was very struck about eighteen months ago when 0:05:42.200 --> 0:05:45.919 several of the senior and early employees of Facebook, people 0:05:45.960 --> 0:05:49.120 like Sean Parker, the first president of Facebook, who have 0:05:49.160 --> 0:05:52.840 now left the firm, obviously Chris Hughes more recently coming 0:05:52.880 --> 0:05:55.800 out and saying, you know, I wouldn't let my children 0:05:56.360 --> 0:06:01.080 use technology. Steve Jobs famously said, um, Steve Jobs, we 0:06:01.120 --> 0:06:03.320 had an audio shoot. I was repeat that. Steve Jobs 0:06:03.360 --> 0:06:06.200 famously gave an interview to Nick Bilton where he said that, 0:06:06.279 --> 0:06:08.240 you know, he wouldn't let his children use the iPad. 0:06:08.880 --> 0:06:10.640 But when the Facebook he's actually just came out and 0:06:10.720 --> 0:06:12.200 one after the other said that they didn't want their 0:06:12.240 --> 0:06:15.040 children using this technology kind of made me sit up 0:06:15.040 --> 0:06:17.799 and think, you know, if the people creating this technology 0:06:18.000 --> 0:06:21.000 don't want their kids to use it, what does that say. 0:06:21.120 --> 0:06:23.080 I mean, it's like, would you go to a restaurant 0:06:23.160 --> 0:06:26.640 where the owner didn't let their children eat? I certainly wouldn't. 0:06:27.000 --> 0:06:29.120 So that was the first sort of point of inspiration 0:06:29.200 --> 0:06:32.400 for for sleepwalkers, in other words, not being aware of 0:06:32.920 --> 0:06:35.640 the future we may be going into. And then there 0:06:35.680 --> 0:06:39.320 was a Zuckerberg hearings in the Senate, and Mark Zuckerberg 0:06:39.400 --> 0:06:45.880 sat there looking increasingly from from slightly nervous two relieved 0:06:45.920 --> 0:06:50.760 and calm to actively smug as it became abundantly clear 0:06:50.960 --> 0:06:52.720 that the senators were not going to be able to 0:06:52.760 --> 0:06:55.560 hold into account. I think the idea of those hearings 0:06:55.880 --> 0:06:59.600 were Senator orn Hatch asking Mark Zuckerberg how the platform 0:06:59.600 --> 0:07:04.120 made my the if it was free and Zugerberg smirkingly replied, Senator, 0:07:04.320 --> 0:07:07.680 we run ads um. And so between those two things, 0:07:07.760 --> 0:07:10.080 between the Facebook is not wanting their own children on 0:07:10.120 --> 0:07:13.400 the platform and the grown ups are either senators not 0:07:13.440 --> 0:07:15.840 being able to hold Facebook to account. I thought, Okay, 0:07:15.840 --> 0:07:18.160 what's going on here? And how can we wake up 0:07:18.160 --> 0:07:20.280 and make sure that we don't sort of flush our 0:07:20.600 --> 0:07:23.800 democracy down the toilet and pollute our children's minds by 0:07:23.800 --> 0:07:26.960 not asking some fundamental questions about how technology is changing 0:07:27.320 --> 0:07:30.600 how we already live. And that brings me to your 0:07:30.600 --> 0:07:33.880 second question, Jonathan, what is AI? And it's a fantastic 0:07:33.960 --> 0:07:37.920 question because AI is everywhere, and it's not just the 0:07:38.000 --> 0:07:41.200 robot future that you see in sci fi films that 0:07:41.200 --> 0:07:44.960 you mentioned, and it's not the future facing products that 0:07:45.000 --> 0:07:48.000 you know many brands tell us they're developing. And it's 0:07:48.040 --> 0:07:51.560 basically just statistics and probability, which has got better and 0:07:51.560 --> 0:07:54.520 better and better over time. But one of the things 0:07:54.560 --> 0:07:56.440 that we make clear to our listeners in the first 0:07:56.440 --> 0:08:00.000 episode is that they've already encountered AI ten times or 0:08:00.040 --> 0:08:02.840 hundred times by the time they listened to this podcast 0:08:02.840 --> 0:08:05.000 in their day, because if they took an uber to 0:08:05.080 --> 0:08:07.440 work in the morning, likely the driver was matched to 0:08:07.520 --> 0:08:10.400 them and the route was chosen with AI. If they 0:08:10.400 --> 0:08:12.280 woke up next to somebody this morning who they met 0:08:12.280 --> 0:08:16.240 through a dating app, AI effectively intervened in their romantic 0:08:16.320 --> 0:08:19.040 life and connected them with somebody who they matched with. 0:08:19.440 --> 0:08:21.920 And even even if you're listening to this podcast right now, 0:08:22.360 --> 0:08:26.000 there are algorithms AI algorithms at work smoothing our voices, 0:08:26.080 --> 0:08:29.840 compressing the audio, helping with the editing techniques. So AI 0:08:29.960 --> 0:08:33.040 is everywhere, and it's already changing our perception of the 0:08:33.080 --> 0:08:35.280 world and how we relate to the world around us 0:08:35.320 --> 0:08:37.680 and each other. Yeah, you could even argue at this 0:08:37.760 --> 0:08:43.920 point that AI is really just a slightly more focused 0:08:44.440 --> 0:08:48.240 branch of computer science and that it's It's almost the 0:08:48.280 --> 0:08:52.920 same as saying will computer science save us or doom us? 0:08:53.080 --> 0:08:55.000 It is too big of a question. You have to 0:08:55.040 --> 0:08:58.800 start narrowing things down. I think the real issue is 0:08:58.840 --> 0:09:03.080 that for the longest time, we've associated artificial intelligence with 0:09:03.160 --> 0:09:07.640 the concept of strong AI, which is that idea that 0:09:07.720 --> 0:09:12.320 we would create a machine that was either capable of 0:09:12.679 --> 0:09:15.760 or so close to capable that we can't tell the 0:09:15.760 --> 0:09:19.240 difference of thinking like a human or processing information like 0:09:19.280 --> 0:09:21.959 a human and coming to decisions like a human would, 0:09:22.000 --> 0:09:29.240 possibly with the added elements of consciousness and self awareness. UM. 0:09:29.360 --> 0:09:33.080 And and you know, I talk about how many times 0:09:33.080 --> 0:09:35.800 in this show. How that's a very complicated thing even 0:09:35.800 --> 0:09:37.880 for us to talk about just as human beings without 0:09:37.880 --> 0:09:44.800 bringing machines into it. So I'm sorry, I can't do that, Jonathan, Yes, yes, 0:09:44.880 --> 0:09:48.920 how HOW or or IBM if you prefer. Uh, you know, 0:09:48.960 --> 0:09:51.920 they're just three letters off UM. But yeah, it's it's 0:09:51.960 --> 0:09:55.400 ah that wonderful, that wonderful feeling that that's the only 0:09:55.440 --> 0:09:58.360 thing that AI really is, right, It's it's the super 0:09:58.440 --> 0:10:04.319 intelligent deep thought or how computer that's capable of processing 0:10:04.320 --> 0:10:09.679 information typically in natural language. Uh, it's the Watson platform 0:10:09.760 --> 0:10:15.000 participating on Jeopardy. Like we've we've precipitated this, uh, this thought, 0:10:15.120 --> 0:10:18.760 this this concept of AI, and we've reinforced it with 0:10:19.720 --> 0:10:23.439 entertainment and with applications that try to emulate the stuff 0:10:23.440 --> 0:10:26.520 that we saw on entertainment. But as you point out, 0:10:26.920 --> 0:10:31.280 AI is is a much more broad concept than this 0:10:31.720 --> 0:10:35.719 super intelligent machine. It's a whole bunch of stuff that's 0:10:35.760 --> 0:10:39.720 all about processing information in a particular way, typically to 0:10:39.920 --> 0:10:44.280 come to some sort of uh, decision or action upon 0:10:44.440 --> 0:10:48.720 information that has been automated. So it might be something 0:10:48.760 --> 0:10:53.120 like Facebook's algorithm, which is all designed ultimately. What Facebook's 0:10:53.120 --> 0:10:55.880 algorithm is designed to do is to keep you on Facebook. 0:10:56.360 --> 0:10:59.959 It's it's ultimately ultimately designed so that you will see 0:11:00.160 --> 0:11:04.160 the next thing on Facebook. It's it's reinforcing that desire 0:11:04.440 --> 0:11:08.560 and uh. And so that's what once the algorithm quote 0:11:08.640 --> 0:11:11.560 unquote figures you out, that's why you're gonna start seeing 0:11:11.800 --> 0:11:17.920 a pretty uh, a pretty consistent presentation of what you 0:11:17.960 --> 0:11:21.280 would see on a day to day basis. UM. But 0:11:21.400 --> 0:11:26.200 that would be one example of that. So, as you 0:11:26.240 --> 0:11:28.720 point out, we do interact with AI all the time, 0:11:28.760 --> 0:11:33.200 whether it's on social media with those algorithms, UM, whether 0:11:33.400 --> 0:11:36.199 it's with an app. Maybe we have one of those 0:11:36.200 --> 0:11:39.560 personal assistants in our home that that uses AI to 0:11:39.760 --> 0:11:43.319 various extents. UM. I talked about just recently on the radio. 0:11:43.360 --> 0:11:47.960 I had a conversation about how Comcast is coming out 0:11:48.080 --> 0:11:52.280 with sensors that are meant to monitor the health of 0:11:52.400 --> 0:11:55.480 people living in homes that have been outfitted with these 0:11:55.520 --> 0:11:58.480 ambient sensors, and they monitor things like how often you 0:11:58.520 --> 0:12:00.800 get up to go to the bathroom or whether you 0:12:00.840 --> 0:12:04.280 stay in bed a longer time than normal. And to 0:12:04.400 --> 0:12:08.160 be perfectly fair to Comcast, they're they're pitching this as 0:12:08.200 --> 0:12:12.319 something to help the elderly or people who otherwise need 0:12:12.440 --> 0:12:16.160 caretakers to give them more independence in their own homes. 0:12:16.200 --> 0:12:20.520 But you could also very easily, without much imagination at all, 0:12:20.880 --> 0:12:23.080 start to come up with scenarios where that could become 0:12:24.320 --> 0:12:27.520 truly invasive. Oh yeah, So I was bringing up Second 0:12:27.600 --> 0:12:30.720 Chance AI, which was a project that came out of 0:12:30.720 --> 0:12:35.840 the University of Washington, which was a was designed to 0:12:36.240 --> 0:12:42.560 detect opioid overdoses early on, using UM an opioid users 0:12:42.640 --> 0:12:46.840 cell phone to detect changes in breath and really act 0:12:46.840 --> 0:12:51.120 as a monitor for people who were long time or 0:12:51.200 --> 0:12:56.000 short time heroin and opioid users. So that device would 0:12:56.080 --> 0:12:59.240 then be able to detect this overdose and allow family 0:12:59.240 --> 0:13:04.640 members to know or also alert the person who is 0:13:04.720 --> 0:13:08.000 overdosing that they're in a bad way. So in the 0:13:08.000 --> 0:13:10.559 case of opioid users, it's worth the trade off because 0:13:11.040 --> 0:13:13.520 um you know, it's very helpful and potentially life saving 0:13:13.559 --> 0:13:16.560 for them to know based on previous breathing patterns and 0:13:16.600 --> 0:13:20.359 previous movements, what's likely to happen next. In an overdose scenario. 0:13:20.520 --> 0:13:22.920 And for most Facebook users, they indeed get to see 0:13:22.920 --> 0:13:25.200 the ads which are relevant to them. But the problem 0:13:25.280 --> 0:13:29.920 with AI is that can't discriminate between individuals and general population. 0:13:29.960 --> 0:13:32.320 So although it's more probable that somebody who have a 0:13:32.320 --> 0:13:35.880 successful pregnancy than not, is very painful for the edge cases, 0:13:35.920 --> 0:13:39.040 and AI can't effectively discriminate for them. And I just 0:13:39.040 --> 0:13:41.120 want to say really quickly, and I think this is 0:13:41.120 --> 0:13:43.280 an important point to make, especially about second Chance. So 0:13:43.360 --> 0:13:47.880 second chance basically is harnessing the power of a cell 0:13:47.920 --> 0:13:51.600 phone's microphone, which is the same microphone that you can 0:13:51.600 --> 0:13:54.800 either choose to turn on or off when you're in Instagram, 0:13:54.840 --> 0:13:57.320 that can listen to what you're saying, and basically then 0:13:57.480 --> 0:14:01.560 use data that's collected to target you with products that 0:14:01.640 --> 0:14:04.440 you probably don't need, like another pair of shoes designed 0:14:04.440 --> 0:14:07.040 by a company that you've never heard of, but that 0:14:07.120 --> 0:14:10.280 you might like. So my point is is that this 0:14:10.320 --> 0:14:14.680 microphone that you know, as Oz was saying, could you 0:14:14.720 --> 0:14:18.040 know in a inappropriately spy on you essentially unless you 0:14:18.120 --> 0:14:20.600 are taking control of it, is also a microphone that 0:14:20.640 --> 0:14:23.000 could save a life of somebody who is in the 0:14:23.000 --> 0:14:25.720 early stages of an opioid overdose. So I think that 0:14:25.800 --> 0:14:28.480 kind of rocks my world when I think about the 0:14:28.520 --> 0:14:33.640 two existing on the same piece of technology. Um, again, 0:14:33.680 --> 0:14:36.160 it's that they're being used for different things, and two 0:14:36.440 --> 0:14:40.440 different things that are you know, have hugely different outcomes. 0:14:40.720 --> 0:14:42.960 But they're all about making guesses about what's going to 0:14:43.000 --> 0:14:45.560 happen in the future based on what's happened in the past, 0:14:45.760 --> 0:14:49.560 and that can be liberating or constraining, depending on the 0:14:49.600 --> 0:14:52.840 technology and the intention and your interaction with it. Yeah, 0:14:53.280 --> 0:14:58.480 I'm reminded of something similar that was it was an 0:14:58.480 --> 0:15:05.600 interesting use of AI that ended up being ah. Another 0:15:06.040 --> 0:15:12.360 embarrassing and emotionally traumatic story that broke a few years ago. 0:15:12.640 --> 0:15:16.480 I want to say it was Target that sent coupons 0:15:16.520 --> 0:15:21.640 like maternity coupons to a young woman who her father 0:15:21.720 --> 0:15:26.640 had intercepted the thing and was incensed that Target would 0:15:26.640 --> 0:15:31.120 send these to his his daughter. Uh. And then because 0:15:31.240 --> 0:15:34.360 the father, the father of the young woman, did not 0:15:34.440 --> 0:15:38.280 realize that she was actually pregnant, she had not told him, 0:15:38.360 --> 0:15:42.320 and so he was upset and he was very angry 0:15:42.360 --> 0:15:45.120 at Target, you know, saying how dare you suggest this? 0:15:45.520 --> 0:15:50.200 Then discovered that she was pregnant after all, and it 0:15:50.280 --> 0:15:54.720 shows again that it was the intent was trying to 0:15:54.720 --> 0:15:56.880 be helpful. You could see that at least from you know, 0:15:56.920 --> 0:15:59.480 from a thousand yards away, you could see that where 0:15:59.800 --> 0:16:02.520 it say a company that says, you know you're going 0:16:02.560 --> 0:16:04.680 to have need of these things, here are some coupons 0:16:04.720 --> 0:16:06.560 for those things. If you shop with us, we can 0:16:06.600 --> 0:16:08.440 get you some deals. So you know it's gonna be 0:16:08.440 --> 0:16:12.200 a mutually beneficial kind of arrangement. But then you realize, oh, 0:16:12.240 --> 0:16:15.320 but this is on a subject that is extremely personal 0:16:15.720 --> 0:16:19.280 and in this case had this unintended consequence. It was 0:16:19.320 --> 0:16:22.400 the same sort of predictive approach, and they were able 0:16:22.440 --> 0:16:25.920 to predict the fact that she was pregnant based upon 0:16:26.160 --> 0:16:31.720 her browsing history. So they were proactively acting on this 0:16:31.960 --> 0:16:36.360 data that had been kind of gathered through her browsing activity. 0:16:36.480 --> 0:16:39.520 And then uh, that's what ended up causing this sort 0:16:39.560 --> 0:16:42.680 of a uh scandal is probably too strong of a 0:16:42.680 --> 0:16:45.400 word for it, but certainly a bruhaha. I think if 0:16:45.920 --> 0:16:49.200 we're looking at the grand scheme of of of how 0:16:49.240 --> 0:16:54.240 do we determine the level of of awkwardness, embarrassment, and 0:16:54.280 --> 0:16:59.520 potential emotional trauma? Um? So yes, please, one of the 0:16:59.560 --> 0:17:02.040 things that can me think about? Jonathan is a study 0:17:02.080 --> 0:17:09.680 at Stanford which basically turned AI onto identifying sexual orientation 0:17:10.000 --> 0:17:13.960 from photographs. So they took a data set publicly available 0:17:14.040 --> 0:17:17.840 data set of images of people's faces from dating websites 0:17:18.200 --> 0:17:22.360 which have been tagged bisexual preference I, straight, gay or bisexual. 0:17:22.840 --> 0:17:26.480 Then they train the algorithm on which faces corresponded to 0:17:26.520 --> 0:17:31.480 which express sexual preferences, and the algorithm, after this training 0:17:31.840 --> 0:17:36.639 was able to identify with accuracy for men and accuracy 0:17:36.800 --> 0:17:41.000 for women sexual orientation just from seeing five photographs of them. 0:17:41.080 --> 0:17:45.640 So again that technology by itself is more or less neutral. 0:17:46.400 --> 0:17:49.879 But you think about it being overlaid onto a citywide 0:17:49.880 --> 0:17:54.160 surveillance system in a country like Brunei or Saudi Arabia 0:17:54.480 --> 0:17:58.520 where homosexuality is punishable up to the death penalty, and 0:17:58.520 --> 0:18:01.240 it starts to become very very scared. Ry. Um. So 0:18:01.440 --> 0:18:03.960 we are in this world now where where technology is 0:18:03.960 --> 0:18:06.720 advancing and the ability to make these predictions based on 0:18:06.800 --> 0:18:09.879 past data is so advanced. It doesn't need to have 0:18:10.000 --> 0:18:15.160 consciousness to be killer, right right, Yeah, The fear of 0:18:15.240 --> 0:18:20.280 the matrix or terminator future, while compelling, turns out to 0:18:20.680 --> 0:18:23.159 not be necessary at all. Like that doesn't need to 0:18:23.160 --> 0:18:26.680 be a component for this to already be dangerous. Yeah, 0:18:26.680 --> 0:18:29.679 we'll we'll go into that in greater detail in just 0:18:30.040 --> 0:18:41.640 a moment, but first let's take a quick break. As 0:18:41.720 --> 0:18:43.399 you were saying just before the break, I mean, you 0:18:43.440 --> 0:18:46.800 made that great point about how AI has this potential 0:18:47.040 --> 0:18:51.239 to do potentially, you know, great harm as a as 0:18:51.240 --> 0:18:53.760 a possibility without the need for any sort of intelligence 0:18:53.880 --> 0:18:56.040 or malevolence on the part of the machine. In fact, 0:18:56.040 --> 0:19:01.280 it can just unthinkingly in human terms, cause some some 0:19:01.359 --> 0:19:07.400 pretty terrible consequences, unintended certainly, or at least we hope so, 0:19:07.880 --> 0:19:10.400 on the on the part of those who designed the systems. 0:19:10.600 --> 0:19:12.280 And I wanted to kind of talk a little bit 0:19:12.320 --> 0:19:15.680 more about that about how sometimes that can happen. And one, 0:19:15.760 --> 0:19:18.679 and I'm sure you've come across this in your reporting 0:19:18.840 --> 0:19:24.040 and in your podcasting. One problem that's not only confined 0:19:24.080 --> 0:19:26.600 to AI, but too and not just a tech but 0:19:26.680 --> 0:19:30.600 across the board is bias. Right, This idea that when 0:19:30.640 --> 0:19:34.720 you're designing a system, you're doing so from a particular 0:19:34.800 --> 0:19:39.160 point of view, and because of that, uh, you are 0:19:39.320 --> 0:19:44.240 likely excluding other points of view, maybe not consciously, but 0:19:44.440 --> 0:19:47.159 you are. And that ends up meaning that if it's 0:19:47.200 --> 0:19:50.399 a system that's supposed to apply to everyone, but it 0:19:51.080 --> 0:19:54.720 particularly applies well to people who are similar to the 0:19:54.720 --> 0:19:57.199 people who designed the system, and not so well to 0:19:57.200 --> 0:20:00.760 everybody else, that becomes a problem. And and we've certainly 0:20:00.760 --> 0:20:05.560 seen this in systems like UM Microsoft Connect. When Microsoft 0:20:05.640 --> 0:20:11.280 was pushing the Connect peripheral, which is the gesture recognition peripheral, 0:20:11.320 --> 0:20:14.320 where there's a camera had an infrared camera and a 0:20:14.359 --> 0:20:18.280 regular optical camera that could detect motions so that it 0:20:18.280 --> 0:20:21.679 could be translated into commands for the system. UM it 0:20:21.760 --> 0:20:24.440 was discovered pretty quickly that it worked great for white 0:20:24.440 --> 0:20:27.639 people but not so great for people of color. It 0:20:28.040 --> 0:20:31.600 had been designed by people who had not really worked 0:20:31.640 --> 0:20:35.439 with it in that regard, and so we see there. 0:20:35.480 --> 0:20:40.000 You could argue a fairly um harmless in the grand 0:20:40.040 --> 0:20:43.160 scheme of things failure of a system, but you look 0:20:43.200 --> 0:20:47.160 at something like computer vision for maybe an autonomous car, 0:20:47.880 --> 0:20:50.000 and you could argue, well, now you're talking about life 0:20:50.080 --> 0:20:53.800 or death situations. So to me, one of the big 0:20:53.920 --> 0:20:58.320 challenges in AI is making sure that you that the 0:20:58.400 --> 0:21:01.840 people designing the systems are doing their best to eliminate 0:21:01.880 --> 0:21:04.760 bias as best they can. And part of that I 0:21:04.800 --> 0:21:12.120 think falls to a real concentrated effort to increase diversity 0:21:12.200 --> 0:21:15.719 in the organization's companies that are actually designing these systems 0:21:15.720 --> 0:21:18.320 in the first place. Yeah. No, absolutely. I mean I 0:21:18.320 --> 0:21:23.320 think that the conversation about AI and bias has sort 0:21:23.320 --> 0:21:27.520 of reached critical critical mass. I guess, you know, I 0:21:27.520 --> 0:21:30.639 think it was yesterday or the day before, you know, UM, 0:21:30.720 --> 0:21:38.280 Alexander Rocascio Cortez was speaking out specifically about this problem 0:21:38.320 --> 0:21:42.000 as it pertains to facial recognition technology. UM, there was 0:21:42.000 --> 0:21:46.160 a very good M I. T study that recently came 0:21:46.160 --> 0:21:48.600 out that you know, a lot of these programs are 0:21:48.640 --> 0:21:54.640 developed by white men and therefore are extremely bias and 0:21:54.440 --> 0:22:00.439 and and I think politicians now are really trying to 0:22:00.480 --> 0:22:04.679 sound the alarm because I think it's, um, it's not 0:22:04.760 --> 0:22:07.159 something people think about in their everyday lives. You know. 0:22:07.560 --> 0:22:09.960 I don't think people are you know, walking around getting 0:22:09.960 --> 0:22:11.640 to their job that maybe they don't want to be at, 0:22:11.760 --> 0:22:13.800 you know, driving to work, driving their kids to school, 0:22:14.000 --> 0:22:16.280 you know, thinking about the implications of bias and facial 0:22:16.320 --> 0:22:19.520 recognition technology. I think people have other things to think about, 0:22:19.520 --> 0:22:24.040 but I think it's very important, UM, especially when you know, 0:22:24.280 --> 0:22:28.520 politicians start bringing up these problems, uh for sort of 0:22:28.640 --> 0:22:31.119 ordinary people to start to think, well, actually, wait a minute, 0:22:31.280 --> 0:22:35.960 I might encounter this technology UM at at border patrol, 0:22:36.160 --> 0:22:38.280 you know, when I'm flying out of the country, or 0:22:38.680 --> 0:22:41.800 you know, I might encounter this technology as I walk 0:22:41.840 --> 0:22:45.639 into a stadium that's now using you know, a quick lane. 0:22:46.119 --> 0:22:50.480 And I think when people start to listen to politicians 0:22:50.520 --> 0:22:54.320 who care about these issues, UM, they realize again that 0:22:54.400 --> 0:22:57.320 there are much more human touch points than we think. 0:22:57.359 --> 0:23:00.359 And then so issues of like bias and gender discrimine nation. 0:23:01.400 --> 0:23:04.600 Whereas before people weren't thinking about them as much in 0:23:04.720 --> 0:23:08.080 terms of technology and artificial intelligence, you know, now people 0:23:08.080 --> 0:23:11.480 are realizing that there's real I don't know, there's there's 0:23:11.520 --> 0:23:14.520 real issues in terms of who is developing these technologies 0:23:14.720 --> 0:23:19.840 and who is harmed by the inherent bias within these technologies. 0:23:20.160 --> 0:23:23.160 And I just want to say something really quickly. One 0:23:23.240 --> 0:23:26.840 hypocrisy that I think is is really wild and worth 0:23:26.880 --> 0:23:30.720 noting is, you know, the European Union has recently released 0:23:30.800 --> 0:23:34.240 basically a list of seven I don't know, I don't 0:23:34.280 --> 0:23:36.400 even know what you call them, but bullet points about 0:23:36.640 --> 0:23:38.960 you know, the way in which we should be talking 0:23:39.000 --> 0:23:43.480 about and regulating artificial intelligence, and you know, one of them, 0:23:43.520 --> 0:23:46.200 one of like the main bullet points is to say, 0:23:46.760 --> 0:23:50.639 you know, we really have to focus on uh by 0:23:50.680 --> 0:23:55.200 the inherent bias UM within these you know, both algorithms 0:23:55.280 --> 0:23:59.440 and the way this technology is built. UM, we don't 0:23:59.480 --> 0:24:01.400 we want to make sure that it doesn't get ahead 0:24:01.440 --> 0:24:04.679 of us essentially, right. And at the same time, the 0:24:04.720 --> 0:24:08.240 European Union in Latvia and Hungary and Greece is using, 0:24:08.359 --> 0:24:12.320 is piloting a program called Eyeborder Control UM, which is 0:24:12.359 --> 0:24:19.439 basically being tested and run by border patrol agents. UM 0:24:19.480 --> 0:24:24.560 two match people's faces on a very very large amount 0:24:24.680 --> 0:24:27.280 of data and then decide if a person should be 0:24:27.320 --> 0:24:32.320 detained for further questioning. Right. So, I think right now, 0:24:33.240 --> 0:24:36.680