WEBVTT - Tech News: You Got AI in my Tech News 0:00:04.360 --> 0:00:12.240 Welcome to tech Stuff, a production from iHeartRadio. Hey there, 0:00:12.320 --> 0:00:15.640 and welcome to tech Stuff. I'm your host, Jonathan Strickland. 0:00:15.640 --> 0:00:18.640 I'm an executive producer with iHeartRadio. And how the tech 0:00:18.720 --> 0:00:22.080 are you. It's time for the tech news for February 0:00:22.239 --> 0:00:26.560 twenty eighth, twenty twenty three. For such a beastly month 0:00:26.600 --> 0:00:30.200 as February twenty eight, days as a rule are plenty. 0:00:31.280 --> 0:00:34.760 Shout out if you recognize that reference. It is another 0:00:34.920 --> 0:00:38.479 AI dominated news day, though I promise there are a 0:00:38.520 --> 0:00:41.519 few stories I'll be covering that do not have AI 0:00:41.560 --> 0:00:45.640 as the central topic. But starting off Elon Musk repartedly 0:00:45.720 --> 0:00:49.280 wants to found a new AI lab to compete against 0:00:49.320 --> 0:00:51.879 open Ai. Now, for those of y'all who remember my 0:00:51.920 --> 0:00:55.680 episode about open ai, maybe this comes as a surprise 0:00:55.800 --> 0:00:58.840 because Musk was actually one of the original co founders 0:00:58.960 --> 0:01:02.080 of open ai in the first place. But back then, 0:01:02.600 --> 0:01:06.039 open ai was a not for profit organization, and it 0:01:06.080 --> 0:01:09.280 had the goal of using an open source approach to 0:01:09.360 --> 0:01:13.000 developing an evolving artificial intelligence in a way that ideally 0:01:13.280 --> 0:01:17.920 would be universally beneficial. You know, none of this AI 0:01:17.959 --> 0:01:21.160 that benefits one group at the expense of everyone else. 0:01:21.319 --> 0:01:25.039 Kind of nonsense. But then Musk stepped down from the 0:01:25.080 --> 0:01:29.160 board of open Ai because ostensibly because Tesla was also 0:01:29.400 --> 0:01:31.600 developing AI of its own and there was a potential 0:01:31.640 --> 0:01:34.800 conflict of interest, though Musk has later said that he 0:01:34.920 --> 0:01:39.400 was critical of open AI's direction, and since that day, 0:01:39.520 --> 0:01:43.479 Musk has gone on to criticize open Ai, particularly once 0:01:43.480 --> 0:01:48.440 the organization founded a for profit arm ostensibly to help 0:01:48.560 --> 0:01:53.320 fund the nonprofit part, and Musk has also criticized chat GPT, 0:01:53.640 --> 0:01:57.960 saying open ai is quote training AI to be woke 0:01:58.600 --> 0:02:02.760 end quote Yeah, Musk, I get it. You're a billionaire 0:02:02.960 --> 0:02:06.480 white guy. Why not punch down because there's nowhere else 0:02:06.520 --> 0:02:09.600 beat a punch right? What a jerk. I'm, of course 0:02:09.600 --> 0:02:12.040 being a little facetious. My opinion of Musk is obviously 0:02:12.080 --> 0:02:15.240 pretty low, but that's beside the point. I know no 0:02:15.280 --> 0:02:17.480 one really wants to hear that, so I'll drop it. 0:02:17.919 --> 0:02:21.079 Musk wants to create this AI lab and pursue AI 0:02:21.160 --> 0:02:25.080 chat butts that are unfettered by the chains of wokeness. 0:02:25.600 --> 0:02:29.440 Considering how Musk has shown that his free speech absolutist 0:02:29.560 --> 0:02:33.440 stance isn't actually in alignment with his behavior, you can 0:02:33.440 --> 0:02:37.520 see also how Twitter had banned mention of competing services 0:02:37.520 --> 0:02:41.720 like massadon Instagram and others on its platform. To see 0:02:41.760 --> 0:02:44.600 evidence of this, I suspect all of this is going 0:02:44.639 --> 0:02:47.160 to come back to haunt him should he actually achieve 0:02:47.240 --> 0:02:51.600 this goal. Meanwhile, Tesla investors are likely further aggravated to 0:02:51.639 --> 0:02:55.320 see that the company's CEO continues to direct his attention 0:02:55.360 --> 0:02:59.959 to yet another endeavor rather than address problems with Tesla. 0:03:00.200 --> 0:03:03.919 And then over at open Ai, the company is introducing 0:03:03.919 --> 0:03:06.840 a platform for developers that will give them access to 0:03:07.080 --> 0:03:12.000 open AI's tools, namely the company's machine learning models. So 0:03:12.200 --> 0:03:17.799 this is the very powerful AI compute systems that to 0:03:17.840 --> 0:03:21.040 build out yourself would require millions and millions of dollars. 0:03:21.720 --> 0:03:26.560 Open ai is calling this offering Foundry, and it means 0:03:26.600 --> 0:03:29.120 that people who have an idea for apps or services 0:03:29.160 --> 0:03:32.680 that would feature AI in some way could have access 0:03:32.760 --> 0:03:36.000 to compute assets without having to build them all themselves. 0:03:36.320 --> 0:03:40.160 That is an enticing offer for developers who might otherwise 0:03:40.160 --> 0:03:43.960 have a great idea but they lack the funds to 0:03:44.040 --> 0:03:47.640 be able to execute upon it. Details are somewhat scarce. 0:03:47.720 --> 0:03:50.360 Open ai has not announced when we might expect Foundry 0:03:50.400 --> 0:03:52.760 to launch, but we do know that it's going to 0:03:52.880 --> 0:03:55.960 set developers back a pretty penny, actually a whole bunch 0:03:56.000 --> 0:03:59.840 of pretty pennies to access these services. According to tech Crunch, 0:04:00.320 --> 0:04:04.640 three months of access to the lightweight version of GPT 0:04:04.640 --> 0:04:08.080 three point five would set you back seventy eight grand. 0:04:08.480 --> 0:04:13.760 That's seventy thousand dollars for three months of access wowsers. 0:04:13.880 --> 0:04:16.320 So this is well beyond the reach of your average 0:04:16.440 --> 0:04:20.360 home developer. We're really talking more about startups and companies 0:04:20.360 --> 0:04:22.840 that have a real shot at seeing a return on investment, 0:04:23.240 --> 0:04:25.600 but they lack the infrastructure or money to build out 0:04:25.640 --> 0:04:30.080 their own machine learning systems. Over at Meta, Mark Zuckerberg 0:04:30.080 --> 0:04:33.560 announced that the company is pursuing its own AI strategy. 0:04:34.000 --> 0:04:37.279 I'm just gonna just read out his Facebook post because 0:04:37.520 --> 0:04:40.760 it tells you everything you need to know. Quote. We're 0:04:40.800 --> 0:04:44.080 creating a new top level product group at Meta focused 0:04:44.080 --> 0:04:48.760 on generative AI to turbo charge our work in this area. 0:04:48.880 --> 0:04:51.880 We're starting by pulling together a lot of teams working 0:04:51.880 --> 0:04:55.880 on generative AI across the company into one group focused 0:04:55.920 --> 0:05:00.880 on building delightful experiences around this technology into all of 0:05:00.880 --> 0:05:04.520 our different products. In the short term, will focus on 0:05:04.600 --> 0:05:08.720 building creative and expressive tools Over the longer term will 0:05:08.720 --> 0:05:12.000 focus on developing AI personas that can help people in 0:05:12.040 --> 0:05:16.279 a variety of ways. We're exploring experiences with text like 0:05:16.480 --> 0:05:21.000 chat in WhatsApp and Messenger, with images like creative Instagram 0:05:21.040 --> 0:05:25.760 filters and ad formats, and with video and multimodal experiences. 0:05:26.120 --> 0:05:28.400 We have a lot of foundational work to do before 0:05:28.440 --> 0:05:32.200 getting to the really futuristic experiences, but I'm excited about 0:05:32.240 --> 0:05:35.640 all the new things will build alone the way end quote. 0:05:36.000 --> 0:05:38.680 So it sounds like Meta like Musk is on the 0:05:38.760 --> 0:05:42.440 path to create its own AI approach, or perhaps Meta 0:05:42.480 --> 0:05:44.719 will turn to Open AI to tap into the power 0:05:44.720 --> 0:05:48.640 of chat GPT. It's still early days and we're not 0:05:48.920 --> 0:05:53.359 done with AI yet. Snapchat is also jumping into the 0:05:53.400 --> 0:05:58.200 AI game with a product it calls my AI. Only 0:05:58.240 --> 0:06:02.600 subscribers to Snapchat Plus will have access to this. According 0:06:02.640 --> 0:06:05.640 to Snapchat, the AI will do stuff like if you 0:06:05.680 --> 0:06:08.320 ask it, it will give you recommendations for presents that 0:06:08.400 --> 0:06:11.840 you could buy friends and family. I mean, presumably Snapchat 0:06:11.880 --> 0:06:14.440 would scan everything you've ever said to these people and 0:06:14.520 --> 0:06:17.640 start to pull suggestions out of that, or it might 0:06:17.680 --> 0:06:21.120 give suggestions about things you could do with somebody, like hey, 0:06:21.279 --> 0:06:23.680 I want to hang out with so and so, what's 0:06:23.720 --> 0:06:25.800 a good activity and they might say, well, they really 0:06:25.800 --> 0:06:28.080 like the outdoors, how about you go hiking that kind 0:06:28.120 --> 0:06:32.400 of thing. You can also apparently name this AI. However, 0:06:32.440 --> 0:06:35.280 the company also owns up to the fact that chat GPT, 0:06:35.920 --> 0:06:41.320 which is the system powering my AI, isn't always reliable. 0:06:41.600 --> 0:06:43.960 We've talked about that a lot over the last couple 0:06:43.960 --> 0:06:47.840 of months, or as Snapchat actually put it, quote, as 0:06:47.920 --> 0:06:52.359 with all aipowered chat bots, my AI is prone to 0:06:52.400 --> 0:06:55.880 hallucination and can be tricked into saying just about anything. 0:06:56.279 --> 0:07:01.159 End quote. Oh. Also, all that communication with AI is 0:07:01.200 --> 0:07:05.440 logged for the purposes of review and development, so anything 0:07:05.480 --> 0:07:09.840 you do say to my AI is being recorded. So 0:07:09.920 --> 0:07:12.680 that means it's best not to confide in my AI 0:07:12.920 --> 0:07:16.720 all your secrets, like Grandma's chocolate chip cookie recipe or 0:07:16.760 --> 0:07:20.720 where you hit the bodies, because someone somewhere could be 0:07:20.760 --> 0:07:26.440 reading over that log. The whole announcement dedicates a surprising 0:07:26.560 --> 0:07:29.320 amount of space that warns users that the tool might 0:07:29.400 --> 0:07:33.360 not work as intended, and that almost raises the question 0:07:33.400 --> 0:07:36.520 about why they're deploying this tool so early. If they're 0:07:36.560 --> 0:07:40.760 taking this amount of effort to say, hey, y'all this 0:07:40.800 --> 0:07:44.640 thing might go heywire, and they really point out all 0:07:44.680 --> 0:07:47.520 the different ways that you can flag stuff so that 0:07:47.560 --> 0:07:51.160 people can review it and thus address any issues that 0:07:51.200 --> 0:07:55.360 pop up. Like it's it's a significant amount of the 0:07:55.400 --> 0:07:59.840 announcement that is all about covering their butts, so to speak. 0:08:00.160 --> 0:08:03.040 So I suppose one answer as to why they're deploying 0:08:03.040 --> 0:08:07.240 it so early is that this turns snapchat plus subscribers 0:08:07.600 --> 0:08:11.280 into QA testers that they don't have to pay. Right, 0:08:11.400 --> 0:08:15.240 these aren't employees. They could turn the community into the 0:08:15.320 --> 0:08:20.680 QA team. It's the basic concept behind open beta programs. Right. 0:08:21.320 --> 0:08:25.640 You find out by using a wide deployment where the 0:08:25.680 --> 0:08:28.400 problems are, and then you fix them. Before you know, 0:08:28.440 --> 0:08:31.040 you deploy it to an even larger audience in the future. 0:08:31.520 --> 0:08:35.440 Jordan Parker Herb wrote a piece for Insider titled I 0:08:35.480 --> 0:08:38.880 asked Chat GPT to write messages to my tender matches. 0:08:39.280 --> 0:08:43.559 A dating coach said they gave off a creepy vibe. Now, 0:08:43.559 --> 0:08:46.439 I don't think anyone's really surprised by that. Heck, if 0:08:46.440 --> 0:08:50.640 you again turned on Nothing Forever the AI powered endless 0:08:50.720 --> 0:08:53.600 Seinfeld episode, you would probably guess this would be the 0:08:53.640 --> 0:08:57.000 inevitable outcome. Because those episodes can get a little unsettling 0:08:57.440 --> 0:09:01.160 as well. And this piece in an Cider indicates that 0:09:01.240 --> 0:09:05.920 chat GPT's responses fell into some pretty common traps when 0:09:06.000 --> 0:09:10.040 one attempts to navigate the complicated world of modern dating, 0:09:10.640 --> 0:09:16.160 namely that chat gpt wrote responses that are way too long. This, 0:09:16.320 --> 0:09:19.199 by the way, it tells me that if I were single, 0:09:19.920 --> 0:09:22.920 I would probably be single forever at this point, because 0:09:23.120 --> 0:09:25.080 come on, y'all, there's no denying. I will use a 0:09:25.160 --> 0:09:28.040 thousand words when ten would do just fine. So I 0:09:28.080 --> 0:09:31.640 would never do well on these kinds of apps. Also, 0:09:31.760 --> 0:09:36.400 chat gpt leaned heavily on its emoji game, and as 0:09:36.480 --> 0:09:38.960 the title of the piece points out, some of the 0:09:39.000 --> 0:09:43.240 responses came across as creepy. Also, the coach pointed out 0:09:43.280 --> 0:09:46.559 that it's best for folks to just be themselves when 0:09:46.640 --> 0:09:49.719 using dating apps because otherwise your perspective date will get 0:09:49.720 --> 0:09:52.400 the wrong impression, and that pretty much guarantees things aren't 0:09:52.440 --> 0:09:56.440 going to go well. Like if a tender matchup thinks 0:09:56.480 --> 0:09:58.880 that the response was really cool, but the really cool 0:09:58.880 --> 0:10:02.719 response was written by AI, and then they meet you 0:10:02.880 --> 0:10:06.520 and you do not have the same vibe. That's a problem. 0:10:06.600 --> 0:10:11.880 I feel like I'm describing almost every romantic comedy that 0:10:12.000 --> 0:10:15.840 was written in the eighties and centered on teenagers, except 0:10:15.920 --> 0:10:18.120 that instead of it being AI, it's typically you know, 0:10:18.160 --> 0:10:22.719 the well meaning popular kids who are attempting to transform 0:10:22.760 --> 0:10:26.120 a person so that they become popular. It feels like that, 0:10:26.240 --> 0:10:30.040 except I guess I'm describing the next generation of teen 0:10:30.200 --> 0:10:33.240 centered comedies. I would not be surprised if we find 0:10:33.240 --> 0:10:36.920 a movie like that. Anyway, I highly recommend reading the 0:10:37.000 --> 0:10:41.040 actual article. Some of the AI generated examples that Jordan's 0:10:41.040 --> 0:10:45.920 shares are absolutely hilarious in that awkward, cringe e sitcom 0:10:46.040 --> 0:10:49.400 kind of way. Again, it's a piece in Insider, and 0:10:49.520 --> 0:10:52.199 it is called I asked chat gpt to write messages 0:10:52.240 --> 0:10:54.400 to my tender matches. You could just search for that. 0:10:54.840 --> 0:10:57.720 I recommend reading it. It's good for, you know, a laugh. 0:10:58.320 --> 0:11:02.080 Of course, AI goes well beyond chatbots and machine learning. 0:11:02.240 --> 0:11:05.080 We've talked about other uses of AI and the dangers 0:11:05.160 --> 0:11:08.840 that they can present. One example that springs to mind 0:11:08.880 --> 0:11:13.000 because it comes up time and again is facial recognition technology. 0:11:13.520 --> 0:11:18.360 Even if the application of this technology is benign, there 0:11:18.360 --> 0:11:22.760 are frequently problems with the underlying tech, unintended bias, has 0:11:22.760 --> 0:11:26.800 been a huge issue with facial recognition technology for years, 0:11:26.920 --> 0:11:30.600 ranging from some services being unable to detect a person 0:11:30.679 --> 0:11:34.760 of colors face properly to misidentification, which can lead to 0:11:34.840 --> 0:11:38.960 traumatic experiences such as being targeted by law enforcement simply 0:11:39.000 --> 0:11:42.880 because a computer can't tell the difference between different people. 0:11:43.400 --> 0:11:48.480 Frequently again people of color, and they are disproportionately targeted 0:11:48.520 --> 0:11:52.319 and affected by such technology. While last week New Scientists 0:11:52.559 --> 0:11:58.160 presented another example, one with truly grim and terrifying implications. 0:11:58.559 --> 0:12:01.560 The magazine found a contract between a tech company called 0:12:01.679 --> 0:12:06.120 real Networks and the United States Air Force. Real Networks 0:12:06.160 --> 0:12:10.400 offers a facial recognition platform that they call Secure Accurate 0:12:10.600 --> 0:12:14.440 Facial Recognition or SAFER, and the implication is that the 0:12:14.480 --> 0:12:17.960 Air Force is incorporating this technology into its Unmanned aerial 0:12:18.120 --> 0:12:22.680 vehicle or UAV program, you know drones. A lack of 0:12:22.720 --> 0:12:25.320 information has led to some speculation, some of which I 0:12:25.320 --> 0:12:30.199 think is definitely understandable and believable. After all, Special Forces 0:12:30.280 --> 0:12:34.199 units have been involved in clandestine operations that are at 0:12:34.280 --> 0:12:39.199 least difficult to separate from stuff like assassinations, and sometimes impossible. 0:12:39.280 --> 0:12:43.280 Sometimes it's just outright an assassination. So it is not 0:12:43.320 --> 0:12:47.080 a stretch to imagine a unit like a Special Forces 0:12:47.160 --> 0:12:51.160 unit making use of a drone with this technology in 0:12:51.240 --> 0:12:55.600 an effort to identify and acquire targets. But the potential 0:12:55.760 --> 0:12:59.200 for misuse of such technology, let alone the chance that 0:12:59.280 --> 0:13:02.640 the techs could make a mistake, has led critics to 0:13:02.720 --> 0:13:05.160 raise the alarm about this approach, and I think that 0:13:05.280 --> 0:13:10.160 is a wise reaction. Even if the tech works perfectly, 0:13:10.840 --> 0:13:12.640 you still have to wrestle with the fact that people 0:13:12.720 --> 0:13:17.319 can sometimes be the absolute worst. They can abuse technology 0:13:17.400 --> 0:13:19.800 for their own purposes, and when it comes to something 0:13:19.840 --> 0:13:25.600 as potentially lethal as this, that is a major problem. Okay, 0:13:25.640 --> 0:13:27.960 we're going to take a quick break. When we come back, 0:13:28.000 --> 0:13:39.960 we'll have a lot more tech news to cover. We're back, 0:13:40.000 --> 0:13:43.240 and we're not done with AI yet. I do promise 0:13:43.280 --> 0:13:45.800 we have other stories besides AI, but we've got a 0:13:45.840 --> 0:13:49.800 couple more to get through, and one of the stories 0:13:49.800 --> 0:13:52.280 we have is about how AI is complicated, not just 0:13:52.360 --> 0:13:56.600 because of the technology, but because of people and the 0:13:56.640 --> 0:14:02.720 way we react to AI and interact with AI. I 0:14:02.760 --> 0:14:07.000 think that this is a truly fascinating topic that relates 0:14:07.000 --> 0:14:11.040 heavily to both psychology and technology. So I want to 0:14:11.040 --> 0:14:13.880 talk about a recent study out of my alma mater, 0:14:14.280 --> 0:14:18.480 the University of Georgia Go Bulldogs. Nicole Davis, who is 0:14:18.480 --> 0:14:22.400 a graduate student at UGA, participated in a research project 0:14:22.800 --> 0:14:26.360 that I think is both interesting and has some upsetting 0:14:26.400 --> 0:14:31.400 but not really surprising conclusions. The project brought together a 0:14:31.480 --> 0:14:34.680 bunch of people and then ask them questions about stereotypes 0:14:34.720 --> 0:14:39.560 that relate to white, Black, and Asian ethnic groups, and generally, 0:14:39.600 --> 0:14:43.760 the response is indicated that people saw Asians as being 0:14:43.800 --> 0:14:48.800 the most competent people and Black people the least competent 0:14:49.160 --> 0:14:54.120 people for any given task. I guess it's a really 0:14:54.200 --> 0:14:58.480 ugly stereotype, but it's also undeniably a pretty common one. 0:14:59.160 --> 0:15:01.720 Then the users were given a task and it was 0:15:01.760 --> 0:15:03.840 to try and find a way to reduce the expense 0:15:03.880 --> 0:15:07.280 of a vacation rental, and they were going to make 0:15:07.440 --> 0:15:11.400 use of an AI powered bot, a chatbot. They had 0:15:11.440 --> 0:15:15.840 a little avatar representing the AI So these were cartoonish 0:15:15.840 --> 0:15:18.680 avatars and there were some that were white, some that 0:15:18.720 --> 0:15:21.880 were black, and some that were Asian in design, and 0:15:22.000 --> 0:15:25.840 the users were later asked to comment on the bot's performance, 0:15:25.960 --> 0:15:30.600 specifically how human and warm it was and how competent 0:15:30.680 --> 0:15:34.480 it was at helping the user reduce the vacation rental cost. 0:15:35.080 --> 0:15:38.640 Davis said, quote when we asked about the bot, we 0:15:38.760 --> 0:15:43.200 saw perceptions change. Even if they said yes, I feel 0:15:43.200 --> 0:15:47.000 like black people are less competent, they also said yes, 0:15:47.440 --> 0:15:51.680 I feel like the black ais were more competent. Davis said. 0:15:51.720 --> 0:15:56.400 This is an example of expectation violation theory, which pausits 0:15:56.440 --> 0:15:59.080 that if someone enters into a situation and they have 0:15:59.200 --> 0:16:02.960 low expectation and then their experience is a positive one, 0:16:03.280 --> 0:16:08.080 they walk away feeling that it was an overwhelmingly positive experience, 0:16:08.760 --> 0:16:12.360 not just that was good, but because it exceeded their expectations, 0:16:12.640 --> 0:16:15.720 it was even better than that. Davis goes on to 0:16:15.760 --> 0:16:17.800 say that more research is needed to find ways in 0:16:17.800 --> 0:16:21.840 which bought representation can help to impact consumer perception in 0:16:21.880 --> 0:16:26.560 positive ways, like perhaps breaking down barriers they might otherwise 0:16:26.640 --> 0:16:30.920 have because of these stereotypes that they maybe unconsciously have 0:16:31.240 --> 0:16:34.920 of different people. But this is obviously a complicated and 0:16:35.080 --> 0:16:39.400 sensitive challenge. Amazon has been using AI to help monitor 0:16:39.440 --> 0:16:42.440 delivery drivers for a while now, but this recently got 0:16:42.480 --> 0:16:45.920 more attention when a TikTok user would the handle. Amber 0:16:46.040 --> 0:16:50.720 Gertz gave an explanation of how the delivery truck's camera 0:16:50.760 --> 0:16:54.720 systems monitor driver behavior. She is an Amazon delivery driver. 0:16:55.160 --> 0:16:58.640 She created this TikTok that explains the whole thing, and 0:16:59.480 --> 0:17:03.480 she says that the system logs violations if a driver 0:17:03.800 --> 0:17:06.879 breaks protocol in any way. This can include stuff like 0:17:07.040 --> 0:17:10.760 failing to come to a complete stop at a stop sign, which, hey, 0:17:10.760 --> 0:17:13.720 that makes sense. This is like one of the biggest 0:17:13.800 --> 0:17:18.320 violations a driver can commit, and you definitely need something 0:17:18.400 --> 0:17:21.479 to help ensure drivers follow this process because I mean, 0:17:21.520 --> 0:17:24.600 they're on the road all day, so they have the 0:17:24.640 --> 0:17:28.439 potential for getting involved in collisions more than the average 0:17:28.480 --> 0:17:30.480 person does. You know, the average person is not on 0:17:30.520 --> 0:17:33.960 the road all day. And it also tracks whether or 0:17:34.000 --> 0:17:36.440 not the driver has buckled their seat belt at the 0:17:36.480 --> 0:17:38.879 conclusion of each stop, or whether or not they've gotten 0:17:38.880 --> 0:17:41.800 out of their seat. However, the system will also trigger 0:17:41.800 --> 0:17:45.120 if a driver takes a drink without first pulling over 0:17:45.200 --> 0:17:46.800 to the side of the road to come to a 0:17:46.840 --> 0:17:50.040 complete stop. So if you got your morning coffee with 0:17:50.119 --> 0:17:52.080 you and you're an Amazon delivery driver, you have to 0:17:52.119 --> 0:17:53.760 come to a complete stop before you can take a 0:17:53.760 --> 0:17:58.000 sip of coffee or you will get you your image 0:17:58.040 --> 0:18:03.879 captured and a violation will be hit on your profile. Also, 0:18:04.480 --> 0:18:07.400 drivers aren't allowed to touch the center console without first 0:18:07.400 --> 0:18:11.040 stopping because that's considered a distraction. And the cameras are 0:18:11.080 --> 0:18:13.760 not providing a live feed for the whole day. It's 0:18:13.760 --> 0:18:17.720 not like there's some security office within Amazon where there's 0:18:17.760 --> 0:18:22.280 this one person looking at a wall of monitors trying 0:18:22.320 --> 0:18:24.280 to keep up with all these different drivers. It would 0:18:24.320 --> 0:18:28.800 be impossible to do that. Instead, AI incorporated into the 0:18:28.840 --> 0:18:32.920 system monitors the camera view and captures video should a 0:18:33.040 --> 0:18:36.320 driver do anything that violates these policies, and it's all 0:18:36.359 --> 0:18:38.720 in the name of safety. As Amber Gert says in 0:18:38.800 --> 0:18:42.200 the TikTok, pretty much every Amazon driver hates this system, 0:18:42.640 --> 0:18:47.080 which includes multiple cameras set up within the vehicle and 0:18:47.160 --> 0:18:50.760 also forward facing cameras to keep things like how far 0:18:50.800 --> 0:18:53.159 away you are from the traffic in front of you. 0:18:53.720 --> 0:18:57.639 But she also generously says this is all in an 0:18:57.680 --> 0:19:02.240 effort to keep drivers and others also Amazon drivers can 0:19:02.320 --> 0:19:05.440 dispute violation reports, and Bergarts even mentions a case where 0:19:05.680 --> 0:19:08.480 a driver scratched his beard while he was driving and 0:19:08.520 --> 0:19:12.160 the system mistakenly believed he was talking on a cell 0:19:12.200 --> 0:19:15.200 phone and so dinged him with a violation, and so 0:19:15.240 --> 0:19:17.560 he was able to dispute that and get it reversed. 0:19:17.600 --> 0:19:20.520 Now I can honestly say I feel really conflicted about 0:19:20.640 --> 0:19:23.399 this whole approach. On the one hand, this is taking 0:19:23.440 --> 0:19:26.800 employee surveillance to the extreme, there's no doubt about it. 0:19:27.240 --> 0:19:29.400 But on the other hand, the system has also allegedly 0:19:29.440 --> 0:19:32.879 contributed to reduction and collision rates by thirty five percent, 0:19:33.320 --> 0:19:37.320 and considering that collisions often result in injuries and property damage, 0:19:37.920 --> 0:19:41.040 that's significant. And I kind of wonder what Ben Franklin 0:19:41.080 --> 0:19:43.520 would have to say about all this, with his views 0:19:43.520 --> 0:19:46.560 on liberty versus safety and all By the way, that 0:19:46.720 --> 0:19:51.520 famous quote is more complicated than the quote itself would indicate. 0:19:51.840 --> 0:19:55.120 I recommend looking into what he was talking about when 0:19:55.119 --> 0:19:58.560 he was chatting about liberty and safety, and hey, I 0:19:58.600 --> 0:20:02.280 mentioned TikTok. Let's talk about that really quickly. Canada has 0:20:02.320 --> 0:20:06.120 now banned TikTok from federal government devices. The White House 0:20:06.119 --> 0:20:07.919 here in the United States has done the same and 0:20:07.960 --> 0:20:12.160 has given federal employees thirty days to wipe TikTok off 0:20:12.200 --> 0:20:15.600 any government owned devices. There are a few special cases 0:20:15.640 --> 0:20:19.119 where there are exceptions for things like security research or 0:20:19.240 --> 0:20:22.879 law enforcement, but for the most part this is a 0:20:23.960 --> 0:20:27.400 federal government wide band. Several state governments in the US 0:20:27.480 --> 0:20:29.960 have done the same sort of thing. The EU has 0:20:30.000 --> 0:20:33.280 started to take action as well for the US and Canada. 0:20:33.359 --> 0:20:36.399 The main concern here is that TikTok's parent company, byte Dance, 0:20:36.520 --> 0:20:39.679 is a Chinese company, and as such could potentially be 0:20:39.760 --> 0:20:42.240 scouring the app for data in an effort to gather 0:20:42.320 --> 0:20:46.760 intelligence on behalf of the Chinese government, specifically the Communist Party. 0:20:47.119 --> 0:20:50.119 For the EU, it gets a little more complicated because 0:20:50.160 --> 0:20:53.480 even if you ignore the connection to China, TikTok itself 0:20:54.000 --> 0:20:56.080 is based in the United States, and the EU is 0:20:56.119 --> 0:20:59.160 a real stickler when it comes to protecting EU citizen 0:20:59.240 --> 0:21:04.639 data from being collected and exploited, and that includes keeping 0:21:04.760 --> 0:21:08.120 the information of the government safe, so they don't want 0:21:08.160 --> 0:21:12.320 the US to just get access to that. Meanwhile, China's 0:21:12.359 --> 0:21:15.840 Foreign Ministry Office issued a statement saying the US quote 0:21:16.280 --> 0:21:19.960 has been overstretching the concept of national security and abusing 0:21:20.000 --> 0:21:24.560 state power to suppress other countries companies. How unsure of itself, 0:21:24.640 --> 0:21:28.800 can the US, the world's top superpower, be to fear 0:21:29.040 --> 0:21:34.760 a favorite young person's favorite app to such a degree 0:21:34.960 --> 0:21:37.760 end quote. First of all, I don't think that favorite 0:21:37.800 --> 0:21:41.200 thing was meant to be repeated, but secondly, shots fired 0:21:41.280 --> 0:21:45.040 China's sick burn. Of course, I should also point out 0:21:45.119 --> 0:21:48.760 that there are literal laws in China that compel citizens 0:21:48.800 --> 0:21:51.320 and companies to act as agents on behalf of gathering 0:21:51.320 --> 0:21:55.960 intelligence for the Communist Party, so there's not a healthy 0:21:56.040 --> 0:22:01.080 leg to stand on there. Also, China, oddly enough, has 0:22:01.240 --> 0:22:05.960 famously blocked tons of apps and services originating in the 0:22:05.960 --> 0:22:09.680 West in an effort to prevent their citizens from accessing them. 0:22:09.680 --> 0:22:13.080 So again, not exactly taking the high ground on that 0:22:13.119 --> 0:22:19.639 front either, but yeah, you use sing us China. Last Pass, 0:22:19.840 --> 0:22:23.240 the password vault company, revealed that hackers were able to 0:22:23.280 --> 0:22:27.320 access and employees home computer and in the process they 0:22:27.400 --> 0:22:31.919 got access to a decrypted vault, a corporate vault, not 0:22:32.000 --> 0:22:35.120 a user vault. This is on the corporate side. Now. 0:22:35.160 --> 0:22:38.320 You might recall that the same service revealed last year 0:22:38.440 --> 0:22:42.440 that hackers had penetrated some customer vaults through other means. Currently, 0:22:42.520 --> 0:22:45.000 last Pass says it does not look like this attack 0:22:45.080 --> 0:22:48.960 and those previous attacks were connected at all. Whatever the case, 0:22:49.320 --> 0:22:53.000 last Pass users should change not only all the passwords 0:22:53.000 --> 0:22:56.200 that they had stored in last passes vault, but also 0:22:56.240 --> 0:22:59.879 their master password for their last pass account. This is 0:23:00.119 --> 0:23:03.960 a worst case scenario, and while Ours Technica points out 0:23:04.280 --> 0:23:07.240 that we do not know yet if hackers have access 0:23:07.280 --> 0:23:11.480 to individual users vaults and their passwords, you have to 0:23:11.520 --> 0:23:15.000 operate under the assumption that they do, and that further, 0:23:15.520 --> 0:23:18.080 this data could end up being sold on the dark web, 0:23:18.160 --> 0:23:20.800 So you definitely want to get out there and start 0:23:20.880 --> 0:23:24.800 changing this stuff now. I've long advocated for password vaults 0:23:24.920 --> 0:23:27.560 as they make the worst parts about passwords a little 0:23:27.600 --> 0:23:30.680 more user friendly. That is, by using a password vault, 0:23:31.080 --> 0:23:35.680 it's easier to create unique, strong passwords for every service 0:23:35.760 --> 0:23:39.520 that you access. These passwords are difficult to crack, but 0:23:39.600 --> 0:23:43.879 they're also hard to remember, and because they're all unique, 0:23:44.240 --> 0:23:48.320 you've got this ton of different passwords that are hard 0:23:48.359 --> 0:23:52.280 for you to just keep in your memory. So it 0:23:52.320 --> 0:23:54.639 gets to the point where it can be impossible to 0:23:54.760 --> 0:23:57.159 remember all of your unique passwords. So a vault's a 0:23:57.160 --> 0:24:01.360 great solution unless something like this happens. And while these 0:24:01.359 --> 0:24:05.119 security events are rare, we've seen they're not impossible and 0:24:05.240 --> 0:24:08.159 that it then falls to us to take action to 0:24:08.200 --> 0:24:11.320 make sure we keep our data and our services as 0:24:11.320 --> 0:24:14.439 safe as possible. Last Pass is not the only target 0:24:14.480 --> 0:24:17.440 to have a catastrophic breach. Another is the United States 0:24:17.560 --> 0:24:21.720 Marshals Service, which announced last week that attackers were able 0:24:21.760 --> 0:24:26.440 to gain access to secure systems or assumed secure systems 0:24:26.440 --> 0:24:29.960 and potentially retrieved sensitive information. The service did say that 0:24:30.000 --> 0:24:33.359 the information may include data about subjects who are currently 0:24:33.480 --> 0:24:39.440 under investigation, It might include administrative information and also personal 0:24:39.520 --> 0:24:42.600 data regarding some of the staff of the agency, among 0:24:42.680 --> 0:24:47.000 other things. However, one system that they said was unaffected 0:24:47.200 --> 0:24:50.520 was the Witness Security Program, which is more commonly known 0:24:50.560 --> 0:24:54.120 here in the US as the Witness Protection Program. This 0:24:54.160 --> 0:24:57.359 is the famous program that aims to create new identities 0:24:57.359 --> 0:25:01.360 for witnesses who are involved in cases for major crimes, 0:25:01.880 --> 0:25:03.920 and this is all in an effort to keep those 0:25:03.960 --> 0:25:08.000 witnesses safe from retaliation. It's pretty much a key ingredient 0:25:08.440 --> 0:25:11.120 in a ton of movies and TV shows that are 0:25:11.200 --> 0:25:15.480 about the mafia. It's frequent that someone gets put into 0:25:15.520 --> 0:25:18.640