WEBVTT - Tech News: Commanding Chatbots to Make Malware 0:00:04.400 --> 0:00:07.800 Welcome to tech Stuff, a production from I Heart Radio. 0:00:11.880 --> 0:00:14.760 Hey there, and welcome to tech Stuff. I'm your host, 0:00:14.960 --> 0:00:18.720 Jonathan Strickland. I'm an executive producer with I Heart Radio. 0:00:19.280 --> 0:00:21.319 And how the tech are you? It's time for the 0:00:21.360 --> 0:00:26.960 tech News for Thursday, January nineteen, twenty twenty three. And 0:00:27.160 --> 0:00:30.160 last year it seemed like the front half of every 0:00:30.200 --> 0:00:34.040 tech news episode was dominated by Elon Musk's on again, 0:00:34.600 --> 0:00:38.920 off again and then totally on again quest to buy Twitter. 0:00:39.640 --> 0:00:43.519 But so far in the honor really goes to chat, 0:00:43.560 --> 0:00:47.120 GPT and open Ai, which is the startup that's behind 0:00:47.159 --> 0:00:52.199 the controversial chat pot so. First up today, Sam Altman, 0:00:52.520 --> 0:00:56.520 the CEO of open Ai, gave an interview with Strictly 0:00:56.720 --> 0:01:00.720 VC in which he addressed some rumors about the upcoming 0:01:00.880 --> 0:01:05.040 next generation of the GPT language model a k a. 0:01:05.760 --> 0:01:11.680 GPT four. We're currently on GPT three point five. Altman 0:01:11.800 --> 0:01:14.920 warned that people are bound to be disappointed in the 0:01:15.040 --> 0:01:18.639 new language model because the hype around it has grown 0:01:18.680 --> 0:01:21.280 to a point that's just impossible to live up to. 0:01:21.840 --> 0:01:24.959 And I think Altman in general has been pretty good 0:01:25.000 --> 0:01:28.200 about acknowledging this kind of stuff. A lot of folks, 0:01:28.240 --> 0:01:32.759 including myself, have pointed out how chat GPT falls short 0:01:32.959 --> 0:01:36.080 in lots of ways, and I don't think Altman would disagree. 0:01:36.600 --> 0:01:40.160 He has been pretty forthcoming about this sort of stuff, 0:01:40.200 --> 0:01:43.560 and he's not making claims that chat GPT is flawless 0:01:43.680 --> 0:01:47.360 or even skilled at every kind of writing. So I 0:01:47.360 --> 0:01:50.600 guess it's unfair for folks like myself to say I 0:01:50.640 --> 0:01:54.120 can't even write a shakespeareance on it, even even though 0:01:54.120 --> 0:01:57.200 it can't. I mean, maybe if I kept at it, 0:01:57.360 --> 0:02:01.200 maybe if I kept tweaking prompts, I could massaw chat 0:02:01.240 --> 0:02:04.280 GPT to a point where it could produce an actual 0:02:04.320 --> 0:02:08.799 Shakespearean style sonnet. But anyway, Altman wouldn't commit to win. 0:02:09.240 --> 0:02:12.560 This next model will launch. He said it will launch 0:02:12.600 --> 0:02:14.840 when it's ready, which I also think it is a 0:02:14.840 --> 0:02:18.200 pretty responsible take, and he admitted that there were concerns 0:02:18.240 --> 0:02:21.440 about how chat GPT might be used by students to 0:02:21.480 --> 0:02:24.720 try and cheat on schoolwork. But he also pointed out 0:02:24.800 --> 0:02:27.680 that when calculators became a thing, we just we didn't 0:02:27.720 --> 0:02:31.120 just abandon teaching students how to do maths. Instead, we 0:02:31.240 --> 0:02:34.640 took calculators kind of into account. While doing so, the 0:02:34.680 --> 0:02:37.600 way we taught it started to change. That's a kind 0:02:37.600 --> 0:02:41.520 of interesting take. Now, Honestly, I think chat GPT could 0:02:41.560 --> 0:02:45.000 be useful for say, English teachers, to show how a 0:02:45.280 --> 0:02:50.440 generated essay might fall short of actual real work done 0:02:50.440 --> 0:02:54.120 by a student who's taking the subject seriously. It could 0:02:54.120 --> 0:02:57.400 be useful to explain to students the difference between real 0:02:57.520 --> 0:03:01.720 critical thinking and surface level as servation. And this way 0:03:01.720 --> 0:03:04.640 you wouldn't have to take an actual student's essay and 0:03:04.680 --> 0:03:06.720 then embarrass that student in front of the rest of 0:03:06.720 --> 0:03:10.080 the class and say, see, little Timmy over there totally 0:03:10.120 --> 0:03:14.480 doesn't understand the role that false staff plays, and Henry 0:03:14.520 --> 0:03:17.400 the fourth part two, just look at this terrible work. 0:03:17.520 --> 0:03:20.560 Shame on you, little Timmy. Instead they could say, look, 0:03:20.639 --> 0:03:24.920 here's this essay written by chat GPT. Here are the 0:03:25.000 --> 0:03:30.040 places where the observations being made are pure surface level 0:03:30.080 --> 0:03:32.480 and they don't really say anything. This is the sort 0:03:32.480 --> 0:03:35.000 of stuff you need to avoid and think about when 0:03:35.000 --> 0:03:38.280 you make your own work. And uh, I get that 0:03:38.320 --> 0:03:41.040 I'm putting a lot on teachers there, and that they're 0:03:41.080 --> 0:03:45.880 already overworked and undervalued, and that should change too, just 0:03:46.040 --> 0:03:49.320 throwing that out there now. One thing about open AI 0:03:49.640 --> 0:03:53.960 that has undeniably caused harm is the process by which 0:03:54.440 --> 0:03:57.840 the company has tried to filter out objectionable material. So 0:03:58.080 --> 0:04:01.840 in the past, even with g PT, we've seen examples 0:04:01.840 --> 0:04:06.120 of how chat bots can submit truly offensive material that 0:04:06.840 --> 0:04:11.280 includes stuff that ranges from racism and sexism and hate 0:04:11.320 --> 0:04:16.159 speech and calls for violence to description of truly horrific 0:04:16.240 --> 0:04:19.599 acts that I can't even begin to describe on this show. 0:04:20.040 --> 0:04:23.760 And obviously, any company making a chat bot wants to 0:04:23.880 --> 0:04:27.880 avoid the chat bot creating these kinds of situations, even 0:04:27.920 --> 0:04:32.040 if it's just because it's bad optics. Right, if the 0:04:32.080 --> 0:04:33.880 reason is, oh, we don't want to do that because 0:04:34.240 --> 0:04:38.760 it'll hurt our investment, at least, they still don't want 0:04:38.839 --> 0:04:42.120 to do it. Now, as you may know, chat GPT 0:04:42.960 --> 0:04:48.320 generates responses to queries by referencing information from a massive 0:04:48.440 --> 0:04:53.200 database of scanned material across the Internet, and there are 0:04:53.279 --> 0:04:57.400 complex rules at play that guide chat GPTs responses, but 0:04:57.480 --> 0:05:03.040 ultimately those responses depend heavily on the repository of scanned information. Now, 0:05:03.040 --> 0:05:06.279 the problem is, and I'm sure you've realized this, the 0:05:06.320 --> 0:05:10.360 web is home to some really terrible communities, ones where 0:05:10.360 --> 0:05:15.320 truly awful material is shared and sometimes celebrated, and chat 0:05:15.360 --> 0:05:20.240 GPT cannot magically tell the difference between what is acceptable 0:05:20.440 --> 0:05:24.600 and what is unacceptable. So it needs people to essentially 0:05:24.640 --> 0:05:28.560 tell the model what isn't isn't right, and it has 0:05:28.600 --> 0:05:31.040 to be able to identify certain stuff as falling into 0:05:31.400 --> 0:05:36.080 categories of content that are forbidden and then filter anything 0:05:36.240 --> 0:05:39.800 like that out of its responses. Anyway, Open a Eyes 0:05:39.839 --> 0:05:44.760 approach was to outsource this work of tagging offensive material, 0:05:44.880 --> 0:05:49.400 identifying it meta tagging it so that chat GPT could 0:05:49.440 --> 0:05:53.000 avoid such stuff, and they went with a company called 0:05:53.080 --> 0:05:56.320 Sama s a m A. This company had also worked 0:05:56.320 --> 0:06:00.880 with Facebook in the area of content moderation, so it 0:06:00.920 --> 0:06:04.680 would hire out people to go through Facebook posts and 0:06:04.800 --> 0:06:09.159 flag any that violated Facebook's policies. Now, the way Sama 0:06:09.160 --> 0:06:12.760 hands handles this is to employ people in Kenya to 0:06:12.839 --> 0:06:15.040 do the work at a salary that breaks down between 0:06:15.080 --> 0:06:18.080 a dollar thirty two and two dollars per hour. By 0:06:18.080 --> 0:06:20.880 the way, Sama was paid around twelve dollars fifty cents 0:06:21.000 --> 0:06:26.640 per hour per case um, and the people actually doing 0:06:26.680 --> 0:06:29.800 the physical work are getting a dollar thirty two to 0:06:29.920 --> 0:06:33.640 two dollars of that time. Magazine indicates that a receptionist 0:06:33.680 --> 0:06:36.920 in Nairobi makes around a dollar fifty two per hour, 0:06:37.080 --> 0:06:40.600 So these are pretty low paying jobs and in return 0:06:41.000 --> 0:06:44.920 for this hourly rate, employees were divided into three groups 0:06:45.360 --> 0:06:50.320 to read passage after passage of just truly the worst 0:06:50.480 --> 0:06:53.680 stuff you can imagine. Each group focused on a specific 0:06:53.760 --> 0:06:56.640 kind of offensive material. There was a hate speech group, 0:06:57.000 --> 0:07:01.000 a sexual abuse group, and a violence group, and I'm 0:07:01.000 --> 0:07:04.240 sure there was plenty of overlap between these three. Employees 0:07:04.240 --> 0:07:07.359 would read and tag each passage for eight or nine 0:07:07.400 --> 0:07:11.080 hours a day, and as you can imagine, being plunged 0:07:11.280 --> 0:07:13.680 into that kind of work and being expected to meet 0:07:14.040 --> 0:07:17.320 certain deliverables right to tag a certain number of passages 0:07:17.360 --> 0:07:21.840 each day really took its toll. It was psychologically taxing, 0:07:21.920 --> 0:07:25.080 to say the least, and many employees reported being traumatized 0:07:25.120 --> 0:07:28.760 by the work, and a few of them argued that SAMA, 0:07:28.960 --> 0:07:34.000 despite what it claimed, was not providing adequate services for counseling. 0:07:34.560 --> 0:07:37.480 And this whole story reminds us that behind the surface 0:07:37.640 --> 0:07:42.440 of these AI programs, there's actually this huge human contribution 0:07:42.520 --> 0:07:46.080 to make them work. Like, yes, it's remarkable that this 0:07:46.160 --> 0:07:50.920 program is composing these texts in response to our queries, 0:07:51.080 --> 0:07:54.840 but to make that possible, a lot of people had 0:07:54.880 --> 0:07:58.320 to spend countless hours of work, and some of them 0:07:58.360 --> 0:08:02.200 doing work that was truly traumatic. The story gets even 0:08:02.280 --> 0:08:07.080 more complicated. Sama recently ended its contracts with open Ai 0:08:07.200 --> 0:08:10.800 ahead of schedule, probably because the company was already being 0:08:10.840 --> 0:08:14.040 taken to task for how they relied on poor people 0:08:14.080 --> 0:08:18.960 in developing countries to moderate content on Facebook, which included 0:08:19.440 --> 0:08:25.720 the content moderators being subjected to truly horrific content, including 0:08:26.160 --> 0:08:29.400 videos and pictures. And we're talking about violence and sexual 0:08:29.440 --> 0:08:34.200 abuse and and and child endangerment and worse, and in 0:08:34.240 --> 0:08:37.319 other words, I don't think Sama made this choice out 0:08:37.360 --> 0:08:42.200 of concern for its workers. The workers found themselves also 0:08:42.320 --> 0:08:45.000 either out of a job entirely or shifted to other 0:08:45.120 --> 0:08:49.160 lower paying work, so they've actually come out poorer for 0:08:49.200 --> 0:08:51.920 this as well. And some of them were saying like, yeah, 0:08:51.960 --> 0:08:54.400 the work is terrible. It takes a toll, but I 0:08:54.480 --> 0:08:57.040 need to provide for my family and now I'm not 0:08:57.120 --> 0:09:00.120 able to. So it's a very grim story. If you 0:09:00.120 --> 0:09:02.640 would like to read about it further, I recommend the 0:09:02.840 --> 0:09:06.280 article in time it is article it is titled rather 0:09:06.760 --> 0:09:11.040 open Ai used Kenyan workers on less than two dollars 0:09:11.040 --> 0:09:15.920 per hour to make chat GPT less toxic. And for 0:09:16.080 --> 0:09:19.560 our third story, about chat GPT. Let's talk about how 0:09:19.600 --> 0:09:22.960 some cybersecurity researchers were able to get the chat bought 0:09:23.080 --> 0:09:26.960 to create a new strand of malware, and not just 0:09:27.040 --> 0:09:31.800 any malware, but polymorphic malware, meaning it can take many forms. 0:09:31.840 --> 0:09:34.800 You can have a core structure that can then be 0:09:34.880 --> 0:09:37.800 tweaked so that you get different generations of malware based 0:09:37.840 --> 0:09:40.560 on the same core, but they can be different enough 0:09:40.840 --> 0:09:44.920 that an anti virus program would have trouble detecting new variants, 0:09:45.320 --> 0:09:48.120 thus extending the useful life of the malware. Like you 0:09:48.120 --> 0:09:51.120 can keep using this core malware over and over by 0:09:51.120 --> 0:09:55.800 making these small changes, relatively small changes, and keep sending 0:09:55.800 --> 0:09:58.440 out waves of the stuff. So this is bad news 0:09:58.520 --> 0:10:00.960 for the security side of things, Lee, but it's also 0:10:01.040 --> 0:10:04.559 important research because if the good guys don't know about it, 0:10:05.080 --> 0:10:07.480 then the bad guys can make greater use of it. 0:10:08.360 --> 0:10:11.280 The researchers said that the web based version of chat 0:10:11.320 --> 0:10:15.640 GPT is a little more challenging to use because it's 0:10:15.679 --> 0:10:18.320 meant to filter out anything that would result in the 0:10:18.360 --> 0:10:21.320 creation of malicious tools. You're not supposed to use chat 0:10:21.360 --> 0:10:23.920 GPT to do it, and it tries to prevent you 0:10:24.080 --> 0:10:27.160 from doing it. However, the team said that if they 0:10:27.200 --> 0:10:30.520 just kept restating their requests and if they made them 0:10:30.520 --> 0:10:34.240 more authoritative, they could eventually work around this barrier and 0:10:34.280 --> 0:10:37.800 convince chat gpt to do their bidding. Which is kind 0:10:37.840 --> 0:10:40.680 of concerning this idea that persistence and a change in 0:10:40.800 --> 0:10:44.680 tone will allow a user to sidestep safety parameters. That's 0:10:44.679 --> 0:10:48.160 not great, But what's arguably worse is the team found 0:10:48.160 --> 0:10:51.920 the a p I, the application programming interface version of 0:10:52.000 --> 0:10:54.280 chat GPT. This is the version that's meant to let 0:10:54.320 --> 0:11:00.000 you incorporate chat GPT functionality into other applications. That version 0:11:00.040 --> 0:11:02.199 appeared to have no such restrictions at play in the 0:11:02.240 --> 0:11:05.839 first place, So, in other words, it was essentially raring 0:11:05.920 --> 0:11:08.760 to go when the time came to make malware. The 0:11:08.840 --> 0:11:12.360 security firm cyber Arc says that it will continue to 0:11:12.600 --> 0:11:15.160 do research on this new development, and that it plans 0:11:15.200 --> 0:11:17.680 to release some of the source code of the malware 0:11:17.760 --> 0:11:21.120 that was created to the cyber security community for the 0:11:21.120 --> 0:11:24.560 purposes of education. And yeah, the bots are learning how 0:11:24.640 --> 0:11:27.600 to create cyber attacks. Now, all right, we're gonna take 0:11:27.600 --> 0:11:29.280 a quick break. When we come back, we'll have some 0:11:29.320 --> 0:11:43.320 more news. We're back now. Last Tuesday, it's past. Tuesday, 0:11:43.320 --> 0:11:45.480 I talked about how crypto is currently in a bit 0:11:45.520 --> 0:11:49.720 of a recovery phase, possibly because the crypto community thinks 0:11:49.760 --> 0:11:54.320 that the macro economic situation is improving. But there's no 0:11:54.440 --> 0:11:57.640 telling if that recovery phase is going to be sustainable 0:11:57.720 --> 0:12:01.440 just yet, and it appears that the big companies are 0:12:01.520 --> 0:12:05.400 still very much in belt tightening mode, which suggests that 0:12:05.600 --> 0:12:08.440 at least these companies expect things to be tough for 0:12:08.480 --> 0:12:11.960 a while. Yet. This week we heard Microsoft is laying 0:12:11.960 --> 0:12:15.240 off around ten thousand employees between now and the end 0:12:15.280 --> 0:12:19.600 of March. A company spokesperson indicated that marketing and sales 0:12:19.679 --> 0:12:23.720 will feel the impact more than engineering, and that gamers 0:12:23.720 --> 0:12:26.440 will also be sad to learn that Xbox and Bethesda 0:12:26.520 --> 0:12:30.320 divisions are among those that will be affected. Satya Nadela, 0:12:30.440 --> 0:12:33.760 the CEO of Microsoft, said that consumers are starting to 0:12:33.760 --> 0:12:38.200 get more frugal with their digital spending and that while 0:12:38.240 --> 0:12:41.240 the early stages of the pandemic saw an increase in 0:12:41.280 --> 0:12:43.880 digital spending, we're now seeing kind of a reversal of 0:12:43.920 --> 0:12:46.920 that trend as people start to ask do I really 0:12:46.960 --> 0:12:50.920 need this digital subscription and start to make choices in 0:12:50.920 --> 0:12:53.920 a way to kind of scale down their own expenditures, 0:12:54.440 --> 0:12:58.000 and on a related note, Amazon continues to cut costs, 0:12:58.160 --> 0:13:02.880 this time by planning to sunset a charitable donation program 0:13:02.920 --> 0:13:06.560 that has been in place since two thousand thirteen. It's 0:13:06.600 --> 0:13:10.120 called Amazon Smile, and it would let customers designate a 0:13:10.200 --> 0:13:14.280 charity one that Amazon had verified, and that charity would 0:13:14.280 --> 0:13:18.080 then receive a percentage of every eligible purchase that the 0:13:18.200 --> 0:13:21.120 customer made on the platform. So instead of going to 0:13:21.400 --> 0:13:25.880 www dot Amazon, you would go to Smile dot Amazon, 0:13:26.280 --> 0:13:29.280 and otherwise everything else would be the same as your 0:13:29.360 --> 0:13:33.240 typical Amazon experience. I actually made use of this program myself. 0:13:33.400 --> 0:13:37.480 I selected a local theater, as in a stage theater 0:13:37.600 --> 0:13:40.480 here in Atlanta to be the recipient because it's a 0:13:40.600 --> 0:13:45.640 nonprofit organization anyway. Now, Amazon is saying that this program 0:13:45.720 --> 0:13:48.400 was not as effective as the company had hoped, and 0:13:48.480 --> 0:13:52.720 that was because these donations were you know, fractions, like 0:13:52.800 --> 0:13:57.800 a percentage of each eligible purchase, and it was spread 0:13:57.840 --> 0:14:03.120 across thousands of charities. No single charity received very much money. 0:14:03.640 --> 0:14:06.000 The company said that the average amount donated to a 0:14:06.120 --> 0:14:10.480 charity amounted to around two d thirty dollars. I just 0:14:10.600 --> 0:14:13.400 checked mine, and over the years that I've been using 0:14:13.400 --> 0:14:15.600 this and I have been using it for several years now. 0:14:16.480 --> 0:14:20.240 My purchases have amounted to about a thousand, seven d 0:14:20.440 --> 0:14:24.080 fifty dollars in donations to this theater. But that's stretched 0:14:24.120 --> 0:14:27.880 across years, right, So it's not like this theater got 0:14:28.000 --> 0:14:31.000 a check for almost two grand and was like, wow, 0:14:31.120 --> 0:14:33.600 what a huge donation. Now it's like, you know, a month, 0:14:33.640 --> 0:14:36.400 they might get a check for you know, a couple 0:14:36.440 --> 0:14:39.960 of bucks. So Amazon kind of has a point here 0:14:40.080 --> 0:14:42.360 in the sense that this was probably not the most 0:14:42.480 --> 0:14:48.000 effective charitable platform. It says that it will still support 0:14:48.080 --> 0:14:52.160 various causes, but it will make more focused decisions on 0:14:52.240 --> 0:14:57.040 things that quote unquote make meaningful change. The Data Privacy 0:14:57.040 --> 0:15:02.440 Commissioner or DPC in Ireland took aim against Meta again, 0:15:03.160 --> 0:15:07.960 this time finding WhatsApp five point five million euro, which 0:15:08.040 --> 0:15:11.800 is just a smidge under six million US dollars, and 0:15:11.880 --> 0:15:15.480 it was for a breach in the EU's strict privacy laws. 0:15:16.240 --> 0:15:20.040 At issue is how WhatsApp has been leveraging the personal 0:15:20.080 --> 0:15:24.840 information of users in the EU while trying to improve 0:15:25.040 --> 0:15:30.040 the company's services. So this isn't about targeted advertising for once. 0:15:30.400 --> 0:15:34.040 It's about using this personal information in a way to 0:15:35.160 --> 0:15:40.720 beef up the services own features and capabilities. Apparently WhatsApp 0:15:40.760 --> 0:15:43.080 methods just didn't go far enough to protect the personal 0:15:43.080 --> 0:15:46.560 information of EU citizens, though the Reuters report I read 0:15:46.640 --> 0:15:49.400 didn't go into really any detail about the nature of 0:15:49.440 --> 0:15:54.000 the violations themselves. Meta has been the target of the 0:15:54.080 --> 0:15:58.360 DPC several times now and has had to pay fines 0:15:58.480 --> 0:16:02.760 on multiple occasions for how and has continued to handle or, 0:16:02.960 --> 0:16:06.360 depending on your point of view, mishandle data pertaining to 0:16:06.640 --> 0:16:12.600 EU citizens. I don't know at what point things would change, 0:16:12.720 --> 0:16:17.120 because the fines, while they're substantial in the big picture 0:16:17.160 --> 0:16:23.680 of Meta's you know, financial situation, rarely amount to anything 0:16:23.720 --> 0:16:28.680 that would raise an eyebrow on on a massive corporate spreadsheet. 0:16:29.360 --> 0:16:33.160 So um, yeah, it's It's yet another example of Meta 0:16:33.240 --> 0:16:36.160 doing what Meta does best, which is handle the personal 0:16:36.200 --> 0:16:41.240 information of its users in a way that's ultimately irresponsible. Meanwhile, 0:16:41.280 --> 0:16:44.360 Global Witness, a human rights watchdog agency that we've talked 0:16:44.360 --> 0:16:48.280 about several times on this show, revealed that Meta's Facebook 0:16:48.600 --> 0:16:53.120 allowed ads calling for violence in Brazil following an already 0:16:53.280 --> 0:16:56.480 violent series of protests in that country. All right, so 0:16:56.560 --> 0:17:00.320 first let's get some background. So in elections late last year, 0:17:00.840 --> 0:17:06.280 the leftist politician Louise Inacio Lula da Silva defeated his 0:17:06.400 --> 0:17:10.760 right wing opponent and former President Bolsonaro in a runoff, 0:17:11.160 --> 0:17:14.680 but Balsonaro took a page from the far far right 0:17:14.800 --> 0:17:19.879 playbook and refused to concede the election. Da Silva took 0:17:19.920 --> 0:17:22.159 office at the beginning of this year, and then some 0:17:22.240 --> 0:17:27.719 of Bosonaro's supporters stormed government buildings in these violent riots 0:17:27.760 --> 0:17:31.600 in an attempt to have this election overturned. The violence 0:17:31.680 --> 0:17:35.199 lasted several hours, and this prompted META to declare Brazil 0:17:35.359 --> 0:17:39.200 a high risk region, at least temporarily. Now that's supposed 0:17:39.200 --> 0:17:43.040 to mean that meta's properties, including Facebook, take a much 0:17:43.040 --> 0:17:47.680 more restrictive approach with regard to content moderation and add approval, 0:17:48.119 --> 0:17:52.240 and an effort to mitigate things like hate speech and 0:17:52.280 --> 0:17:55.639 calls for violence. But Global Witness found that just days 0:17:55.800 --> 0:18:01.680 after the riots, Facebook accepted fourteen of sixteen fake ads 0:18:01.720 --> 0:18:06.200 that Global Witness created, and these ads called for violence 0:18:06.240 --> 0:18:10.199 against the Silva and his supporters. That is definitely not 0:18:10.359 --> 0:18:13.840 a good look. Facebook should not have accepted those ads, 0:18:13.840 --> 0:18:17.480 but it was totally prepared to run fourteen of sixteen 0:18:17.520 --> 0:18:20.680 of them. Global Witness reps say that this shows how 0:18:20.760 --> 0:18:24.000 Meta fails to live up to its obligations in these 0:18:24.080 --> 0:18:26.600 kinds of cases, and it's something we've seen play out 0:18:26.760 --> 0:18:31.720 around the world, particularly in non English speaking countries. There's 0:18:31.800 --> 0:18:35.440 long been this accusation against Meta that the company pays 0:18:35.560 --> 0:18:41.520 very little attention to content moderation outside of major English 0:18:41.640 --> 0:18:45.240 speaking markets like the United States and the EU, and 0:18:45.359 --> 0:18:49.600 yet by doing this, by ignoring these other regions, the 0:18:49.640 --> 0:18:54.400 company is facilitating serious harm to come to populations by 0:18:54.760 --> 0:18:57.879 not cracking down on things like hate speech and calls 0:18:57.920 --> 0:19:02.119 for violence. Global Witness says that YouTube performed substantially better 0:19:02.280 --> 0:19:05.640 than Meta did and denied ads that had similar language 0:19:05.640 --> 0:19:09.200 and messaging in them, and just to set minds at ease, 0:19:09.440 --> 0:19:13.160 Global Witness did cancel these ads before they could actually 0:19:13.200 --> 0:19:18.359 go live, so no users encountered these ads that violated 0:19:18.400 --> 0:19:22.920 Facebook's policies and called for violence because Global Witness pulled 0:19:22.960 --> 0:19:26.840 them before they could they could go live on the site. Okay, 0:19:26.880 --> 0:19:30.119 I've got a couple more stories to get to, but 0:19:30.359 --> 0:19:32.639 before I get to those, let's take another quick break. 0:19:42.400 --> 0:19:47.200 We're back. So I've got a story about video games. Uh, 0:19:47.440 --> 0:19:50.080 it's not necessarily a positive one. So one of the 0:19:50.119 --> 0:19:54.879 ongoing issues in the world of video games involves what 0:19:55.080 --> 0:19:58.080 comes along with leaning on the games as a service 0:19:58.240 --> 0:20:02.320 business model, that is, releasing a game that has elements 0:20:02.359 --> 0:20:04.920 in it that allow a company to continue to generate 0:20:05.000 --> 0:20:10.399 revenue from players over time. So in the old old days, 0:20:10.520 --> 0:20:12.080 the way you made money if you were in the 0:20:12.160 --> 0:20:15.240 video game businesses, you sold as many copies of a 0:20:15.320 --> 0:20:18.760 video game as you possibly could. But once you sold 0:20:18.800 --> 0:20:21.000 a copy of a video game, that was kind of it. 0:20:21.160 --> 0:20:24.119 That was the end of the transaction. Then once you 0:20:24.200 --> 0:20:28.200 started getting into games that had a subscription model like 0:20:28.600 --> 0:20:32.159 M M O RPGs, there became this new way to 0:20:32.240 --> 0:20:35.120 make money. First you'd sell the game, then you would 0:20:35.119 --> 0:20:39.080 continue to make money from the game by collecting subscription fees. 0:20:39.680 --> 0:20:43.800 And this started to open up possibilities where you know, 0:20:43.840 --> 0:20:46.080 you look at this and you think, all right, we 0:20:46.119 --> 0:20:49.520 can continue to make profit from these games. We do 0:20:49.640 --> 0:20:53.040 have to also put effort in to continue to support 0:20:53.080 --> 0:20:55.280 the game, So there is an ongoing cost as well. 0:20:55.320 --> 0:20:58.119 It's not like, you know, we work one day and 0:20:58.160 --> 0:21:00.119 then we just make passive income for the rest or 0:21:00.119 --> 0:21:04.160 our lives. That's that's a myth, but it did open 0:21:04.240 --> 0:21:06.359 up these opportunities, and that's when we started to see 0:21:06.440 --> 0:21:13.600 other ways to make money on ongoing titles, and that 0:21:13.760 --> 0:21:18.080 involved some stuff like being able to pay for upgrades, 0:21:18.160 --> 0:21:21.040 cosmetic upgrades for your character so that they could look 0:21:21.520 --> 0:21:24.879 a specific way and have a specific style. And most gamers, 0:21:24.960 --> 0:21:27.600 I don't think object to this, like the idea of yeah, 0:21:27.720 --> 0:21:29.280 you know, if you want, you can pay a dollar 0:21:29.520 --> 0:21:32.480 and you get this new skin for your character and 0:21:32.600 --> 0:21:35.920 it looks really cool, and I think most gamers are like, yeah, 0:21:36.000 --> 0:21:38.199 I can, you know, I don't care about that, so 0:21:38.280 --> 0:21:40.480 I'm not going to spend the money, or yeah it's 0:21:40.480 --> 0:21:42.480 a dollar, I'll spend a dollar and help support this 0:21:42.560 --> 0:21:45.360 game and plus get this cool skin. Most people don't 0:21:45.400 --> 0:21:48.800 have objections to that. But then you have the darker 0:21:48.880 --> 0:21:53.480 side of this, where you know, they offer up items 0:21:53.640 --> 0:21:58.360 or outfits or just you know, elements. They give players 0:21:58.400 --> 0:22:01.879 an advantage in return for some cold hard cash that 0:22:01.960 --> 0:22:06.800 you buy something that boosts your character performance, either temporarily 0:22:07.000 --> 0:22:10.280 or permanently, and a lot of players refer to this 0:22:10.400 --> 0:22:14.280 as the pay to win style right where you you 0:22:14.280 --> 0:22:18.640 can get an advantage on better players just by spending 0:22:18.640 --> 0:22:22.239 money and it it has a really bad rep in 0:22:22.320 --> 0:22:26.480 the video game world. Generally speaking, it's it's heavily frowned 0:22:26.520 --> 0:22:29.040 upon because people who put in the time and who 0:22:29.080 --> 0:22:33.160 genuinely love the game will find themselves frustrated with folks 0:22:33.200 --> 0:22:36.520 who just have more discretionary income and who are spending 0:22:36.560 --> 0:22:39.080 more on the title in order to get the advantage. 0:22:39.119 --> 0:22:42.360 And that's very frustrating, because we deal with that enough 0:22:42.400 --> 0:22:45.400 in real life, y'all. But anyway, along with these monetization 0:22:45.440 --> 0:22:50.520 strategies come concerns that the games, you know, gamify stuff 0:22:50.680 --> 0:22:54.320 so that players are kind of lured into spending money 0:22:54.359 --> 0:22:57.479 that otherwise they wouldn't, or there might be elements of 0:22:57.600 --> 0:23:01.800 gambling at play, such as the purchase of a lootbox. 0:23:02.119 --> 0:23:07.240 So a lootbox doesn't guarantee you a specific item within 0:23:07.280 --> 0:23:12.120 the game. It gives you a chance of getting different items, 0:23:12.119 --> 0:23:15.560 and the more common items tend to have a larger chance, 0:23:15.600 --> 0:23:18.000 and the more rare items tend to have a smaller chance, 0:23:18.040 --> 0:23:21.520 So it becomes kind of like a gamble. And a 0:23:21.560 --> 0:23:24.720 lot of countries have looked into this and even gone 0:23:24.760 --> 0:23:28.560 so far as to call it gambling and outlawed loot boxes. Well, 0:23:28.560 --> 0:23:30.840 now the EU is kind of looking in on this too, 0:23:31.960 --> 0:23:36.320 as well as other elements of of games as a 0:23:36.320 --> 0:23:40.200 service that could have a negative impact on gamer's mental health. 0:23:40.400 --> 0:23:44.080 So this doesn't mean that the EU is going to 0:23:44.200 --> 0:23:48.840 vote against loot boxes or or make those you know, illegal, 0:23:48.960 --> 0:23:51.800 or regulate them in any way. We haven't gotten to 0:23:51.840 --> 0:23:55.199 that step yet, but they have adopted this report that 0:23:55.359 --> 0:23:58.680 is looking into these kinds of things, which could be 0:23:59.119 --> 0:24:03.040 the first step toward such an outcome. So I imagine 0:24:03.080 --> 0:24:06.280 there are some video game companies out there that are 0:24:06.320 --> 0:24:11.480 watching this with some anticipation because if the EU does 0:24:12.040 --> 0:24:15.000 come to the conclusion that stuff like lootboxes amount to 0:24:15.080 --> 0:24:17.399 gambling and that this has to be regulated, and that 0:24:17.560 --> 0:24:21.400 maybe that also creates restrictions on who can buy the games, 0:24:21.840 --> 0:24:25.840 that's going to have a massive impact of the industry. Um. Yeah, 0:24:25.880 --> 0:24:29.760 So it's it's again an ongoing issue that is still 0:24:29.840 --> 0:24:32.760 kind of shaking out in different places around the world. 0:24:33.560 --> 0:24:37.080 Maybe in five or ten years we're going to see 0:24:37.160 --> 0:24:43.000 dramatically different strategies for monetization to avoid these kinds of 0:24:44.280 --> 0:24:47.600 I guess what Obi wan Kenobi would call imperial entanglements, 0:24:47.640 --> 0:24:49.879 although human in a bad way, whereas I think like, 0:24:50.480 --> 0:24:52.320 maybe we do need to rain this in a bit, 0:24:52.359 --> 0:24:56.160 because I think it is get a bit predatory. And finally, 0:24:56.720 --> 0:25:00.000 a group of academic researchers have proposed that social network 0:25:00.440 --> 0:25:05.040 should employ a new approach to their recommendation algorithms. These 0:25:05.080 --> 0:25:09.119 are the sets of rules that determine what content to 0:25:09.240 --> 0:25:13.320 promote to each user at any given time. Now, it's 0:25:13.320 --> 0:25:17.720 hardly controversial to say that, as of right now, recommendation 0:25:17.760 --> 0:25:23.040 algorithms tend to promote harmful material, or at least material 0:25:23.119 --> 0:25:26.480 that is more likely to create larger divisions between different 0:25:26.480 --> 0:25:30.600 populations of people, such as people who have different political leanings. 0:25:31.440 --> 0:25:35.480 The purpose of the algorithm is not too so discord. 0:25:35.600 --> 0:25:38.879 It's not to watch the world burn, it's not to 0:25:38.960 --> 0:25:43.919 create chaos. The purpose for the algorithm is simply to 0:25:44.160 --> 0:25:48.640 keep users engaged, and more importantly, to keep them on 0:25:48.880 --> 0:25:53.280 the respective platform for as long as possible. Keep them 0:25:53.320 --> 0:25:56.960