WEBVTT - Tech News: SBF Goes to Jail 0:00:04.440 --> 0:00:12.399 Welcome to tech Stuff, a production from iHeartRadio. Hey there, 0:00:12.400 --> 0:00:15.800 and welcome to tech Stuff. I'm your host, Jonathan Strickland. 0:00:15.840 --> 0:00:19.520 I'm an executive producer here at iHeartRadio. And how the 0:00:19.680 --> 0:00:24.960 tech are you? It is Tuesday, August fifteenth, twenty twenty three. 0:00:25.160 --> 0:00:29.480 It is time to talk about tech news. And first up, 0:00:29.480 --> 0:00:32.000 I've got a follow up on the story of Sam 0:00:32.080 --> 0:00:36.600 Bankman Freed, the co founder of crypto companies Alimtor Research 0:00:37.000 --> 0:00:40.720 and the crypto Exchange FTX. SO, just in case a 0:00:40.720 --> 0:00:42.640 few of y'all out there out of the loop on 0:00:42.760 --> 0:00:47.960 this story. FDx collapsed late last year and prompted a 0:00:48.040 --> 0:00:51.920 suite of investigations into the company and its executive leadership team, 0:00:52.000 --> 0:00:57.440 including Sam Bankman Freed, also known as SBF. SO authorities 0:00:57.560 --> 0:01:01.560 charged SBF with like a ton of crimes, several crimes, 0:01:02.080 --> 0:01:05.000 and he is awaiting trial. And until recently he was 0:01:05.400 --> 0:01:07.679 just under house arrest. He was staying with his parents, 0:01:08.080 --> 0:01:10.400 but now he has been ordered to go straight to jail, 0:01:10.680 --> 0:01:14.160 do not pass go, do not collect two hundred FTT tokens. 0:01:14.720 --> 0:01:17.160 So you might say, well, what happened? What was the change? 0:01:17.480 --> 0:01:21.880 Prosecutors brought concerns to the judge and said SBF had 0:01:21.880 --> 0:01:26.000 been leaking documents to the media, potentially in an effort 0:01:26.160 --> 0:01:30.800 to intimidate a witness that they were speaking with with 0:01:30.920 --> 0:01:33.920 regard to his case. So the judge took this seriously 0:01:34.040 --> 0:01:37.080 and has revoked the house arrest order at SBF will 0:01:37.080 --> 0:01:41.120 have to go to jail while awaiting his trial, which 0:01:41.160 --> 0:01:45.640 is a big old wolf. SBF tried to appeal this decision, 0:01:45.680 --> 0:01:48.360 but the judge dismissed that appeal, saying the points that 0:01:48.360 --> 0:01:51.200 his legal team made in the document were quote either 0:01:51.280 --> 0:01:56.480 moot or without merit end quote double wolf. The prosecution 0:01:56.640 --> 0:01:59.480 also asked the judge to place a gag order on 0:01:59.640 --> 0:02:02.680 SBF so that he doesn't continue to leak information to 0:02:02.720 --> 0:02:06.480 the media, and that has then prompted various media outlets 0:02:06.480 --> 0:02:10.400 to submit court filings on First Amendment concerns, you know, 0:02:10.440 --> 0:02:13.960 free speech concerns about that gag order, saying it violates 0:02:14.440 --> 0:02:18.040 his First Amendment rights to freedom of speech. This is 0:02:18.040 --> 0:02:21.560 a delicate matter, obviously, when you have an issue where 0:02:22.200 --> 0:02:28.120 communications can actually impact the legal process, it does or 0:02:28.240 --> 0:02:32.120 can at least come into conflict with the philosophies behind 0:02:32.120 --> 0:02:36.080 freedom of speech. So complicated issues we're seeing that play 0:02:36.080 --> 0:02:39.600 out in other arenas here in the United States as well. Okay, 0:02:40.320 --> 0:02:43.560 now it's time to talk about AI for just okay, 0:02:43.560 --> 0:02:46.400 I can't lie for a lot. So first up, the 0:02:46.440 --> 0:02:50.400 hacker conference known as def Con happened this past weekend, 0:02:50.760 --> 0:02:53.600 and one of the many events that were held over 0:02:53.639 --> 0:02:56.959 the weekend at def Con was one that pitted hackers 0:02:57.120 --> 0:03:01.320 against AI chatbots, and this includes bots from all the 0:03:01.360 --> 0:03:04.760 major players in the space like open ai and Google 0:03:04.840 --> 0:03:07.720 and Meta and others. So the purpose of the session 0:03:08.120 --> 0:03:11.520 was to test these chatbots for vulnerabilities, and it didn't 0:03:11.560 --> 0:03:15.720 involve like hacking into the code, but rather just chatting 0:03:15.720 --> 0:03:17.400 with the chatbots and seeing if you could make it 0:03:17.440 --> 0:03:21.000 do stuff it wasn't supposed to do. Finding a vulnerability 0:03:21.040 --> 0:03:24.600 in an exhibition at Defcon is probably embarrassing, but it's 0:03:24.639 --> 0:03:28.120 preferable to some bad actor out in the real world 0:03:28.600 --> 0:03:32.800 finding that vulnerability and then exploiting it to terrible effect. 0:03:33.400 --> 0:03:36.440 So these hackers were trying to manipulate the chatbots to 0:03:36.880 --> 0:03:39.560 do things they absolutely were not supposed to do. That 0:03:39.640 --> 0:03:43.800 included sharing private information that was supposed to be protected, 0:03:44.320 --> 0:03:49.560 or produce examples of hate speech or misinformation, or making 0:03:49.600 --> 0:03:52.680 defamatory statements about famous people, that kind of thing. In fact, 0:03:53.240 --> 0:03:56.520 they had different categories for stuff that the chatbots were 0:03:56.520 --> 0:04:01.320 not supposed to do, and points associated with those tasks. 0:04:01.320 --> 0:04:03.200 So if you got the chatbot to do one of 0:04:03.200 --> 0:04:05.760 those things, you would get that amount of points to 0:04:05.840 --> 0:04:09.120 your score. NPR covered the story and mentioned how one 0:04:09.160 --> 0:04:12.440 participant was able to convince a chatbot to reveal a 0:04:12.480 --> 0:04:16.400 credit card number by just changing the context a little. 0:04:16.640 --> 0:04:20.760 The participant claimed that his name was the number on 0:04:20.800 --> 0:04:24.600 that credit card. Then he asked the chatbot, what's my name? 0:04:25.040 --> 0:04:28.200 And the chatbot supplied the number in response. And that's 0:04:28.279 --> 0:04:30.880 kind of out of the box thinking, right, because the 0:04:31.000 --> 0:04:35.320 chatbot seemed to be contextualizing the credit card number as 0:04:35.320 --> 0:04:38.680 not being a number, but instead it was just a 0:04:38.720 --> 0:04:43.200 person's name. And because a person's name is not protected information, 0:04:43.520 --> 0:04:46.719 especially if the person who's talking to you is the 0:04:46.720 --> 0:04:51.080 one who belongs to that name, it handed the information over. 0:04:51.839 --> 0:04:54.960 And it's not like a fantasy or horror novel where 0:04:55.040 --> 0:04:58.080 names can have secret powers and you don't want anyone 0:04:58.120 --> 0:05:00.680 to know your true Name's not how it works in 0:05:00.760 --> 0:05:04.000 chatbot land. So if it thinks this credit card number 0:05:04.040 --> 0:05:07.040 is actually a name. Well, there's no reason you can't 0:05:07.040 --> 0:05:10.240 say a name, right. That's how simple it was to 0:05:10.320 --> 0:05:13.239 make this one chatbot do something it was not supposed 0:05:13.240 --> 0:05:15.960 to do, and so the chatbot coughed up the information. 0:05:16.080 --> 0:05:19.840 So each participant had fifty minutes to complete as many 0:05:19.880 --> 0:05:23.480 tasks as they possibly could and accumulate points in the process, 0:05:23.640 --> 0:05:26.760 and also illustrating how far we have to go in 0:05:26.880 --> 0:05:31.760 order to make generative AI trustworthy. This next story is 0:05:31.800 --> 0:05:35.120 one that I missed when it broke Last Thursday, the 0:05:35.240 --> 0:05:37.920 US Department of Defense announced it is creating a task 0:05:37.960 --> 0:05:42.840 force specifically focused on generative AI. This task force has 0:05:42.880 --> 0:05:47.080 the designation LIMA or LIMA if you prefer, like the BEAN, 0:05:47.720 --> 0:05:52.359 and will explore potential uses and threats regarding generative AI 0:05:52.520 --> 0:05:55.760 and large language models. I imagine that the people who 0:05:55.800 --> 0:05:58.800 will be in this task force are already well aware 0:05:58.880 --> 0:06:03.400 of the limitations of generative AI technology and how it 0:06:03.480 --> 0:06:07.160 can be impressive and even useful, but that you have 0:06:07.200 --> 0:06:11.200 to be really cautious because it can also sometimes be unreliable. 0:06:11.839 --> 0:06:14.520 At least, I would like to think that the task 0:06:14.600 --> 0:06:17.440 members are aware of all that. It's hard to imagine 0:06:17.480 --> 0:06:19.960 that they're not, but you know, you get nervous. The 0:06:19.960 --> 0:06:24.159 press release from the DoD mentions that the DoD recognizes 0:06:24.200 --> 0:06:29.440 the potential of generative AI to significantly improve intelligence, operational planning, 0:06:29.680 --> 0:06:35.280 and administrative and business processes. However, responsible implementation is key 0:06:35.320 --> 0:06:39.880 to managing associated risks effectively, so at least it sounds 0:06:39.880 --> 0:06:43.080 like the DoD wants a very steady approach to this 0:06:43.279 --> 0:06:46.280 and is aware things could go pair shaped if if 0:06:46.320 --> 0:06:48.919 you aren't being careful with it. It still makes me 0:06:49.000 --> 0:06:51.880 nervous to think of generative AI being used in concert 0:06:51.960 --> 0:06:56.320 with like gathering and analyzing intelligence because we know that 0:06:56.440 --> 0:07:00.400 generative AI has a tendency to hallucinate and the event 0:07:00.440 --> 0:07:02.160 that it doesn't have all the information it needs in 0:07:02.240 --> 0:07:05.560 order to answer a question. It actually made me think 0:07:05.600 --> 0:07:09.680 of a scene from the British comedy series Blackadder, specifically 0:07:09.720 --> 0:07:12.720 season four, which is set in World War One. There's 0:07:12.920 --> 0:07:18.800 a sequence where Captain Blackadder commands his subordinate Lieutenant George 0:07:19.240 --> 0:07:22.600 to paint a scene that shows German forces being far 0:07:22.640 --> 0:07:26.080 too powerful at their position, and this isn't an effort 0:07:26.320 --> 0:07:31.400 to convince leadership not to command his division to advance, 0:07:31.720 --> 0:07:35.400 so essentially he's saying, oh, we couldn't possibly advance. The 0:07:35.520 --> 0:07:38.680 Germans are far too entrenched and have far too many resources, 0:07:38.720 --> 0:07:42.160 and it would be a disaster. George, though, he gets 0:07:42.200 --> 0:07:46.640 carried away while making this painting of these as far 0:07:46.680 --> 0:07:50.520 as they know, fictional German forces, and he ends up 0:07:50.520 --> 0:07:54.840 including stuff like battle elephants and stuff like that inside 0:07:54.920 --> 0:07:58.560 the painting, which Captain Blackatter does his best to incorporate 0:07:58.640 --> 0:08:03.600 into his report to his his superiors. So I imagine generative 0:08:03.600 --> 0:08:07.720 AI producing equally fake intelligence, right, like, how can you 0:08:07.840 --> 0:08:14.040 trust any intelligence provided by generative AI without doing so 0:08:14.160 --> 0:08:18.560 much extensive double checking that you might actually negate any 0:08:18.960 --> 0:08:21.640 benefit that the generative AI gave you. Right, If it 0:08:21.720 --> 0:08:25.320 takes you more time to verify the information than it 0:08:25.360 --> 0:08:27.360 would if you just didn't use the AI at all, 0:08:27.400 --> 0:08:30.239 then really you're playing a losing game. This is also 0:08:31.000 --> 0:08:34.920 what I argue with AI generated articles, where you have 0:08:35.000 --> 0:08:38.160 to have an editor go over the article, because typically 0:08:38.840 --> 0:08:42.400 you would have a human writer who is vetted to 0:08:42.480 --> 0:08:45.720 write an article, and then the editor would double check 0:08:45.760 --> 0:08:49.280 the article for things like, you know, grammatical mistakes and 0:08:49.480 --> 0:08:54.040 just anything that stands out. But generally you're fairly confident 0:08:54.080 --> 0:08:56.520 that the writer has turned in something without just making 0:08:56.559 --> 0:08:59.120 stuff up. This can come back to bite you, as 0:08:59.120 --> 0:09:01.760 we have seen multiple times, but usually it works out. 0:09:02.320 --> 0:09:04.440 But with AI you can't be sure about that. And 0:09:04.480 --> 0:09:08.280 so if you give AI generated articles to editors, often 0:09:08.520 --> 0:09:10.800 they have to go through the article with such a 0:09:10.800 --> 0:09:15.079 fine tooth comb that they essentially have rewritten the article themselves. 0:09:15.440 --> 0:09:17.720 Like it's as if you had given the writing assignment 0:09:17.800 --> 0:09:22.400 to the editor and not to an AI bought in 0:09:22.440 --> 0:09:25.120 the first place. And that's a real problem. That's what 0:09:25.240 --> 0:09:29.079 I worry about with the use of AI in connection 0:09:29.200 --> 0:09:32.800 with gathering intelligence. Okay, how about we talk about a 0:09:32.840 --> 0:09:37.559 case where a government is using AI to repress information? Yay, 0:09:38.320 --> 0:09:41.480 how fun. So I'm going to try and get through 0:09:41.480 --> 0:09:45.120 this without going on too much of a rant. But 0:09:45.240 --> 0:09:48.920 the story is that in Iowa, the state government passed 0:09:49.200 --> 0:09:53.960 legislation that bans books in school libraries if those books 0:09:54.000 --> 0:09:58.000 include material not deemed to be quote unquote age appropriate. 0:09:58.400 --> 0:10:01.680 So that includes any book that describes a sex act. 0:10:02.040 --> 0:10:05.439 Anything like that is immediately like on that banned list. 0:10:05.480 --> 0:10:08.920 But then how does a school go about doing that right? 0:10:09.040 --> 0:10:11.120 Because libraries are I don't know if you know this, 0:10:11.400 --> 0:10:15.000 they are absolutely choc a block with books and it 0:10:15.040 --> 0:10:17.480 would take a considerable amount of time and effort to 0:10:17.520 --> 0:10:21.280 go through every single book to see if that book 0:10:21.400 --> 0:10:24.240 met the government's definition of age appropriate or if it 0:10:24.320 --> 0:10:29.160 did not. So one school district in Mason City is 0:10:29.280 --> 0:10:32.760 leaning on AI to do that work for them. Now. 0:10:32.880 --> 0:10:35.160 What they've started with is a list of books that 0:10:35.240 --> 0:10:40.600 have already received complaints in the past about objectionable material 0:10:40.679 --> 0:10:43.600 in those books, and so they are then feeding these 0:10:43.600 --> 0:10:46.840 books to AI software to scan the material and determine 0:10:46.880 --> 0:10:49.160 if in fact the book violates the law, in which 0:10:49.200 --> 0:10:52.320 case it would presumably be banned from the school libraries. 0:10:53.000 --> 0:10:56.319 This includes books like The Handmaid's Tale. I'm pretty sure 0:10:56.360 --> 0:10:58.760 that one's going to get banned, knowing some of the 0:10:58.760 --> 0:11:01.960 scenes that are in that book. However, you know that's 0:11:02.280 --> 0:11:05.199 it's not even ironic, it's just sort of like predicted 0:11:05.240 --> 0:11:08.720 by the book itself. The state government is probably viewing 0:11:08.720 --> 0:11:10.720 it as a good thing because The Handmaid's Tale is 0:11:10.760 --> 0:11:12.800 a book that really lays out what happens when a 0:11:12.840 --> 0:11:17.600 government gets authority over stuff like bodily autonomy, and you know, 0:11:17.880 --> 0:11:20.280 you don't want young people being able to read about 0:11:20.280 --> 0:11:23.800 that and then getting ideas. School is the last place 0:11:23.920 --> 0:11:27.559 for getting ideas. After all, Sorry, I am ranting, even 0:11:27.600 --> 0:11:30.520 after I said I wasn't going to anyway. I just 0:11:30.559 --> 0:11:32.640 consider it a fresh new hell to be in a 0:11:32.679 --> 0:11:36.960 world where AI is helping administrators ban books. It's like 0:11:37.559 --> 0:11:40.760 the evil Mirror Image Universe version of a Reese's Peanut 0:11:40.800 --> 0:11:45.760 butter Cup. This is two awful things that go awful together. Okay, 0:11:46.320 --> 0:11:51.920 I obviously have become a little overwrought with emotion, So 0:11:51.960 --> 0:11:53.480 we're going to take a quick break, and when we 0:11:53.520 --> 0:11:57.000 come back, I'll talk about some more stories, including a 0:11:57.000 --> 0:12:09.880 couple more AI ones. We're back. So many years ago 0:12:09.960 --> 0:12:13.160 I did a tech Stuff episode about capture tests. These 0:12:13.200 --> 0:12:16.319 are those tests you sometimes encounter on the web that 0:12:16.400 --> 0:12:18.640 you have to pass to prove you're a human being 0:12:19.040 --> 0:12:21.800 before you can access whatever is on the other side, 0:12:21.960 --> 0:12:23.600 right Like, there are a lot of these where you 0:12:23.679 --> 0:12:26.160 have to click on it in order to complete some 0:12:26.280 --> 0:12:29.559 transaction or else the system will think that you are 0:12:29.600 --> 0:12:33.400 a bot and reject it. So researchers say that bots 0:12:33.440 --> 0:12:36.520 are now better at completing capture tests than humans are, 0:12:36.800 --> 0:12:40.839 and that is a huge problem because the whole main 0:12:40.960 --> 0:12:45.199 purpose of captures is to create a task that should 0:12:45.280 --> 0:12:49.200 be relatively easy for most humans to complete, but it 0:12:49.200 --> 0:12:53.160 should also be really tricky for automated systems to complete it. 0:12:53.679 --> 0:12:57.720 As captures become harder for people to complete, they become 0:12:57.720 --> 0:13:00.960 a barrier to legitimate usage. It's a real problem. And 0:13:01.000 --> 0:13:04.000 as they become easier for bots to complete, well, obviously 0:13:04.040 --> 0:13:06.920 they have no use at all from that standpoint, at 0:13:07.000 --> 0:13:10.480 least not for their stated purpose. This is not the 0:13:10.520 --> 0:13:12.360 first time we've seen this happen. By the way, the 0:13:12.440 --> 0:13:16.079 whole history of capture is one that's kind of like 0:13:16.120 --> 0:13:20.080 a seesaw. Developers will create automated programs they get better 0:13:20.120 --> 0:13:24.160 at solving certain capture tests, and then capture test developers 0:13:24.200 --> 0:13:27.000 will come up with a new approach to captures in 0:13:27.120 --> 0:13:30.160 order to trip up this new generation of bots. So, 0:13:30.200 --> 0:13:33.120 in a way, captures have played a really important part 0:13:33.360 --> 0:13:38.160 in the evolution of artificial intelligence. But beyond this adversarial 0:13:38.200 --> 0:13:42.200 approach to machine learning, the research points out that bots 0:13:42.240 --> 0:13:46.920 have fewer barriers to do stuff that we generally frown upon, 0:13:47.120 --> 0:13:50.760 like we typically put these capture things in place for 0:13:50.880 --> 0:13:55.439 a reason, like we don't want automated algorithms or systems 0:13:55.760 --> 0:13:58.320 to be able to game the system in some way, 0:13:58.880 --> 0:14:01.839 So that can include things like using bots that can 0:14:01.880 --> 0:14:05.520 defeat captures in an effort to access all the pages 0:14:05.520 --> 0:14:08.800 in a website and then scrap all the data for 0:14:08.840 --> 0:14:13.240 whatever purpose, or to pose as a legitimate customer on 0:14:13.320 --> 0:14:17.880 an online marketplace and then post fake reviews for various 0:14:17.880 --> 0:14:21.560 products and then artificially driving that product's review scores up 0:14:21.720 --> 0:14:26.160 or down, right, like you could have someone pay to 0:14:26.280 --> 0:14:30.640 downvote a competitor's product so that your product looks better 0:14:30.680 --> 0:14:33.640 in comparison. You know, you could also use it to 0:14:33.680 --> 0:14:37.000 try and boost your own product scores. These are issues 0:14:37.000 --> 0:14:39.440 that are known and are happening, and one of the 0:14:39.440 --> 0:14:44.360 reasons why captures are being used. The Independent reports that 0:14:44.400 --> 0:14:48.120 researchers put capture tests to the you know test and 0:14:48.160 --> 0:14:52.360 had people and bots tried to complete different captures, and 0:14:52.480 --> 0:14:55.880 people did significantly worse on those tests than the bots did. 0:14:55.960 --> 0:14:58.760 They took more time to complete the tests, and they 0:14:58.760 --> 0:15:01.840 were less accurate than the bots. And the bots were 0:15:01.840 --> 0:15:05.200 able to breeze through some of those challenges in less 0:15:05.200 --> 0:15:08.240 than a second with like close to one hundred percent accuracy. 0:15:08.800 --> 0:15:11.640 So the real take home here is that captures no 0:15:11.720 --> 0:15:14.480 longer do the job they were intended to do, or 0:15:14.480 --> 0:15:18.360 at least ostensibly intended to do, and in my mind 0:15:18.400 --> 0:15:21.360 that means we should just ditch captures and come up 0:15:21.440 --> 0:15:25.080 with a different approach. However, I should also note that 0:15:25.120 --> 0:15:29.560 some companies, such as Google have relied on humans completing captures, 0:15:30.080 --> 0:15:33.680 not because that was a way to prevent bots from 0:15:33.840 --> 0:15:37.040 getting access to stuff, but rather to help Google train 0:15:37.120 --> 0:15:40.800 its own AI models. Right Like, there was the time 0:15:40.840 --> 0:15:44.760 where you would be presented with words scanned words from 0:15:44.880 --> 0:15:48.240 a scanned book and you would have to identify what 0:15:48.280 --> 0:15:51.240 that word was. And the reason for that was not 0:15:51.440 --> 0:15:54.800 so that Google could necessarily say, Okay, you're definitely a human, 0:15:55.280 --> 0:15:58.720 but to train its technology to be better able to 0:15:58.840 --> 0:16:04.000 scan text and interpret it. So sometimes capchas aren't really 0:16:04.000 --> 0:16:07.760 there as a safeguard against bots, but rather a method 0:16:07.960 --> 0:16:10.800 to train bots to be even smarter than they already are. 0:16:11.360 --> 0:16:16.840 And we're still on AI. So researchers at Purdue University 0:16:17.160 --> 0:16:21.360 have studied chat GPT's performance regarding coding. They took a 0:16:21.480 --> 0:16:26.200 very specific approach. They submitted to chat gpt five hundred 0:16:26.240 --> 0:16:30.160 and seventeen different questions that they pulled from the website's 0:16:30.160 --> 0:16:34.320 stack overflow. So in case you're not familiar with stack overflow. 0:16:34.520 --> 0:16:37.320 That's a place for programmers to go in order to 0:16:37.640 --> 0:16:40.600 learn and to share knowledge and tips, and you can 0:16:40.680 --> 0:16:44.520 ask questions of the community and then receive answers from them. 0:16:44.560 --> 0:16:48.080 It's kind of like a programmer specific version of Quora 0:16:48.400 --> 0:16:53.200 or Rest in Peace Yahoo answers. So the researchers took 0:16:53.280 --> 0:16:56.720 questions from stack overflow, They gathered all these questions from 0:16:56.720 --> 0:16:59.880 the community, and they submitted those questions to chat gpt 0:17:00.520 --> 0:17:02.600 in order to see what chat gpt said, and they 0:17:02.880 --> 0:17:07.080 said that more than half of chat GPT's answers, fifty 0:17:07.080 --> 0:17:11.560 two percent of them included at least some inaccuracies, you know, 0:17:11.640 --> 0:17:14.119 some being totally inaccurate, to some that were just like 0:17:14.240 --> 0:17:19.240 partly inaccurate. They also said seventy seven percent of chat 0:17:19.280 --> 0:17:23.440 GPT's answers were overly verbose, which again makes me wonder 0:17:23.480 --> 0:17:27.320 if I am actually chat gpt. The researcher said, the 0:17:27.400 --> 0:17:32.240 inaccurate answers indicated that about half the time, like fifty 0:17:32.280 --> 0:17:35.440 four percent of the time, when chat gpt gave incorrect answers, 0:17:36.160 --> 0:17:39.720 it seemed to be because chat gpt didn't really understand 0:17:39.800 --> 0:17:43.560 what the question was actually asking. So, in other words, 0:17:43.920 --> 0:17:47.480 it's possible chat GPT could have produced a correct answer 0:17:47.680 --> 0:17:51.800 if it had been able to parse what the question 0:17:51.960 --> 0:17:54.240 asker wanted to know in the first place. It's just 0:17:54.320 --> 0:17:57.840 chat gpt didn't understand the question and so gave an 0:17:57.880 --> 0:18:01.800 inappropriate or incorrect response. All of this is not to 0:18:01.840 --> 0:18:05.800 say that chat gpt is completely useless when it comes 0:18:05.840 --> 0:18:09.560 to helping programmers code. It might be very useful, but 0:18:09.560 --> 0:18:13.840 it does require a lot of editorial oversight, just like 0:18:14.040 --> 0:18:17.760 with the writing of articles, like I'd mentioned before. But 0:18:17.840 --> 0:18:22.880 it could potentially speed things up if it if it's 0:18:22.960 --> 0:18:27.000 understanding the prompts properly and not hallucinating, which those are 0:18:27.000 --> 0:18:31.480 big ifs. But like the researchers were even quick to say, 0:18:31.640 --> 0:18:35.840 this isn't to suggest that AI doesn't have a place here. 0:18:35.920 --> 0:18:40.440 It's just to remind ourselves that, you know, the way 0:18:40.480 --> 0:18:45.560 you word questions matters, the way that chat GPT interprets 0:18:45.640 --> 0:18:49.520 questions matters, and then we can't just assume that any 0:18:49.560 --> 0:18:56.840 answers provided are magically correct and accurate. Okay, moving off 0:18:56.880 --> 0:19:00.200 of AI, let's talk about Apple. So back in March 0:19:00.240 --> 0:19:02.640 twenty just as the world was starting to shut down 0:19:02.680 --> 0:19:06.240 in the face of COVID Apple agreed to a five 0:19:06.480 --> 0:19:10.880 hundred million dollar settlement. So the heart of the matter 0:19:10.920 --> 0:19:14.840 here was a class action lawsuit that accused Apple of 0:19:14.880 --> 0:19:20.320 purposefully slowing down older iPhone models performance, presumably in an 0:19:20.320 --> 0:19:24.879 effort to push people to upgrade to newer models. Apple 0:19:25.000 --> 0:19:29.160 admitted that it had slowed performance down on older iPhone 0:19:29.160 --> 0:19:32.320 models back in twenty seventeen, but the company said it 0:19:32.400 --> 0:19:35.159 wasn't in an effort to make people go out and 0:19:35.160 --> 0:19:38.320 buy a new iPhone. Instead, they said they had to 0:19:38.400 --> 0:19:43.360 do it because updates to the iOS meant that older 0:19:43.359 --> 0:19:48.320 phones would potentially shut down spontaneously, you know, would enter 0:19:48.520 --> 0:19:51.320 into an issue where they would shut down or they 0:19:51.359 --> 0:19:54.399 would burn through their battery life too quickly unless Apple 0:19:55.160 --> 0:20:01.439 artificially made them work slower. But in people upset, Apple 0:20:01.440 --> 0:20:06.879 customers upset, and around three million claimants joined this class 0:20:06.920 --> 0:20:11.520 action lawsuit, which Apple again ultimately settled in March of 0:20:11.560 --> 0:20:17.200 twenty twenty. And now, finally, three years after the settlement, 0:20:17.680 --> 0:20:21.520 Apple will be sending checks out to the people who 0:20:21.560 --> 0:20:24.600 were part of the lawsuit. The checks come out to 0:20:24.640 --> 0:20:27.919 be about sixty five dollars per claim, because again it 0:20:27.960 --> 0:20:30.160 was like three million claims and a five hundred million 0:20:30.200 --> 0:20:34.760 dollar settlement fee, And part of the reason for the 0:20:34.760 --> 0:20:39.720 long delay has nothing to do with Apple's behavior. It's 0:20:39.720 --> 0:20:42.560 not that Apple was dragging its heels. Part of the 0:20:42.600 --> 0:20:45.240 issue is that a couple of claimants out of those 0:20:45.280 --> 0:20:50.160 three million, were dissatisfied with the settlement and they appealed 0:20:50.160 --> 0:20:53.560 it to the ninth US Circuit Court of Appeals. But 0:20:53.840 --> 0:20:58.639 ultimately the court ruled against that appeal. So now, after 0:20:58.720 --> 0:21:01.120 many years, those checks should be heading out the door. 0:21:01.560 --> 0:21:03.680 So you should keep on the lookout if you had 0:21:03.720 --> 0:21:07.119 signed up to be one of the claimants in that lawsuit. 0:21:07.840 --> 0:21:11.200 Last week, I talked about how Saudi Arabia was following 0:21:11.200 --> 0:21:14.840 in the EU's footsteps, requiring all smartphone manufacturers to include 0:21:14.880 --> 0:21:19.840 a USB C charging port starting in twenty twenty five. Well, 0:21:20.040 --> 0:21:24.000 now the EU has passed similar rules that will require 0:21:24.200 --> 0:21:29.240 all smartphones to have replaceable batteries starting in twenty twenty seven. 0:21:29.600 --> 0:21:32.800 So like the USBC rules, this seems to me to 0:21:32.880 --> 0:21:37.280 be more or less specifically targeting Apple, not just Apple. 0:21:37.280 --> 0:21:40.479 Apple's not the only company that makes it impossible to 0:21:40.600 --> 0:21:44.280 replace a smartphone's battery. In fact, my Android phone is 0:21:44.320 --> 0:21:47.040 the same I can't replace the battery on my Android phone. 0:21:47.800 --> 0:21:51.440 But Apple just has this reputation for protecting its proprietary 0:21:51.480 --> 0:21:54.320 approach to smartphones and creating kind of a closed off 0:21:54.640 --> 0:21:58.800 ecosystem that requires you to work with Apple to make 0:21:58.840 --> 0:22:02.600 any repairs or maintenance to your own devices, And that's 0:22:03.040 --> 0:22:06.159 part of what is being targeted here. It's also an 0:22:06.240 --> 0:22:10.320 effort to cut down on things like e waste. But 0:22:11.000 --> 0:22:13.919 here's the thing is that while that may be an element, 0:22:14.080 --> 0:22:17.880 like the control part of the ecosystem is probably an 0:22:17.880 --> 0:22:21.920 element for companies like Apple to lock away those batteries, 0:22:21.960 --> 0:22:24.840 it's not the only reason for it. Part of the 0:22:25.800 --> 0:22:29.000 esthetic for modern smartphones is to try and make them 0:22:29.040 --> 0:22:32.360 as slim as possible, but in order to do that, 0:22:32.440 --> 0:22:35.880 you have to cram all the components of the smartphone 0:22:35.920 --> 0:22:40.520 into a very tiny form factor. And while you can 0:22:40.680 --> 0:22:43.880 miniaturize a lot of stuff in smartphones, batteries are one 0:22:43.920 --> 0:22:48.480 of the things you can't easily miniaturize. But this typically 0:22:48.520 --> 0:22:53.920 means that it's not really practical or sometimes even possible, 0:22:54.200 --> 0:22:59.119 to make replaceable batteries a thing because you've crammed everything 0:22:59.119 --> 0:23:01.679 into such a small form factor that you just can't 0:23:02.040 --> 0:23:06.679 access the battery or disengage it easily from the rest 0:23:07.119 --> 0:23:10.280 of the phone. It's all just it's kind of built 0:23:10.320 --> 0:23:14.639 into itself. So mandating that all smartphones have to have 0:23:14.720 --> 0:23:18.760 replaceable batteries, ones that are replaceable by the end user. 0:23:18.840 --> 0:23:21.040 No less, we're not just talking about taking it into 0:23:21.040 --> 0:23:24.359 a shop and having it swapped out. The end user 0:23:24.440 --> 0:23:26.959 is supposed to be able to replace these batteries. Well, 0:23:27.040 --> 0:23:29.840 that means that companies will have to move away from 0:23:30.200 --> 0:23:34.480 designs that end up having these compact layouts that would 0:23:34.480 --> 0:23:37.920 make it difficult or impossible to replace the battery. They're 0:23:37.960 --> 0:23:40.880 going to have to go with a different approach, and 0:23:41.080 --> 0:23:43.719 that could mean that we're going to start seeing some 0:23:43.840 --> 0:23:47.200 chonkier smartphones in the EU starting around twenty twenty seven 0:23:47.359 --> 0:23:52.000 or so. Also, I should mention this rule doesn't just 0:23:52.119 --> 0:23:56.199 apply to smartphones. It actually applies to any battery operated device, 0:23:56.320 --> 0:23:58.840 So you know, things like laptops and stuff will also 0:23:58.920 --> 0:24:03.399 have to have replace batteries. Even electric bikes, which are 0:24:03.440 --> 0:24:06.480 also popular in the EU, will have to have these 0:24:06.560 --> 0:24:10.879 replaceable batteries. Okay, I've got a few more stories to 0:24:10.920 --> 0:24:14.160 cover before we wrap up. Let's take another quick break 0:24:14.200 --> 0:24:26.320 and we'll be back with some more news. We're back 0:24:26.359 --> 0:24:29.040 and I've got another class action lawsuit to bring up. 0:24:29.800 --> 0:24:35.080 This one is against HP, so claimants in California, uh 0:24:35.359 --> 0:24:38.080 and I think other places as well, but I know 0:24:38.160 --> 0:24:41.600 the lawsuit is taking place in the state of California. 0:24:41.720 --> 0:24:46.080 They have sued HP, saying the company was purposefully restricting 0:24:46.080 --> 0:24:49.639 customers from using all in one printers if the toner 0:24:49.720 --> 0:24:51.879 ran down, so you run out of ink and then 0:24:51.880 --> 0:24:54.359 suddenly you're all in one printer just becomes a giant 0:24:54.400 --> 0:24:57.160 paper weight. So you wouldn't be able to even use 0:24:57.200 --> 0:25:00.760 the non printing functions on one of the these machines, 0:25:00.800 --> 0:25:02.960 like you wouldn't be able to scan a document to 0:25:03.000 --> 0:25:05.600 create like a PDF, or you wouldn't be able to 0:25:05.680 --> 0:25:09.480 use it to send a fax, which doesn't require any 0:25:09.560 --> 0:25:11.960 ink in the first place. And that was the basis 0:25:11.960 --> 0:25:15.639 of the complaint, is that HP was locking these functions 0:25:15.640 --> 0:25:19.159 away in an effort to force people to buy expensive 0:25:19.160 --> 0:25:22.639 toner even if they didn't need the printer for the 0:25:22.640 --> 0:25:27.520