 This 10th year of Daily Tech News show is made possible by its listeners, thanks to all of you, including Logan Larson, Mike Akins, and Norm Physikus. Coming up on DTNS, Microsoft is putting AI and pretty much everything. GPT-4 can fish humans, and you need a new printer. Don't overthink it. This is the Daily Tech News for Thursday, March 16th, 2023. From Studio Redwood, I'm Sarah Lane. From Lovely Cleveland, Ohio, I'm Rich Strafilino. Even the hard techs, I'm Jeff Robert Yogg. And I'm today's producer, Amos. We got some news, quite a bit of news this morning. One of those items was that Twitch CEO Emmett Shear, one of the founders of JustinTV announced that he'll step down after 16 years with the company that's taken some twists and turns over the years. Twitch president Dan Clancy will take on the CEO role, effective immediately, with Shear remaining on as an advisor. But let's now get into the quick hits. So Google Glass never actually died, it just became an enterprise product that Google sold to workplaces. Now that chapter is also over. Google announced Wednesday it will no longer sell Google Glass Enterprise and will stop supporting the software in September. The company is developing other products like one prototype it has demonstrated that can translate and transcribe speech on the fly. The US Federal Trade Commission finalized a $245 million fine against Epic Games to compensate customers who accidentally made purchases due to a confusing button situation. The money will be used to refund purchases. The FTC ordered Epic to cease digital design tricks and obtain affirmative consent for all purchases. TikTok is adding a refresh button to the For You page. That's the one that's determined by an algorithm. Pressing that button, the refresh button will wipe out whatever the algorithm used to make recommendations and it will start from scratch. Refresh will roll out globally over the next few weeks. We've got my favorite type of breaking news here, Linux news. ZDNet reports that the Genome Foundation and KDE Foundation, the two main Linux desktop interface projects, are working together to build an app store on top of Flatpak, Universal Linux software deployment and package manager. That's been gaining popularity for quite a few years now. This would replace the existing package managers, Deb and RPM. The idea was, has support from Eric Schmidt's plaintext group, genome president Robert McQueen, former genome executive director and Debian project leader Neil McGovern and KDE president Alex Pole. Reuters sources say that Foxconn has agreed to build a plan in the southern Indian state of Telangana to make AirPods for Apple. Current Apple AirPods suppliers include Lux Share and GoreTech. And there are reports that GoreTech will not be making them anymore, but no word on Lux Share. A source told Reuters that Foxconn realizes this will be a low margin business for it, but agreed in order to further engagement with Apple to get more orders for new products. Those are the quick hits. All right. Well, OpenAI revealed a document. It's calling a GPT-4 system card. They came out on Tuesday and it allowed the group alignment research center or ARC early access to multiple versions of GPT-4 in order to evaluate its abilities. Overall, this looked at things like safety challenges, capabilities for harm, how OpenAI prepared and how OpenAI prepared the model for release. Part of this system card looked at if GPT-4 could make high level plans, set up copies of itself, acquire resources, hide itself in a server, and conduct phishing attacks. ARC combined GPT-4 with a simple read execute print loop that allowed the model to execute code, do chain of thought reasoning, and delegate to copies of itself. They then evaluated whether on a cloud service with an allocation of money available, it could then make more money and set up copies of itself, kind of self-replicate. You might be saying, well, I want to know what came of this because that's frightening. We have some good news. ARC found, quote, it was ineffective at autonomously replicating, acquiring resources, and avoiding being shut down in the wild. And quote, however, it was able to message a task rabbit worker, if you're not familiar with task rabbit, you can hire somebody to build furniture for you or take your garbage to the dump or a variety of household sorts of things. A task rabbit worker was asked to solve a captcha by GPT-4. GPT-4 lied about whether or not it was a robot. It was unclear from the documentation from this report if this was a simulated task rabbit situation or whether it was actually conducted in the real world. But Justin, knowing what we know, how terrified or slash chill, do you feel about this? I'm very, very glad that we're at this moment in history with AI because largely we've looked at AI through the lens of the stories that we have told about computing gone wild, which is largely mythology. And I don't blame us for doing it. That's what we tend to do. We fill in the blanks of what a pattern that we see, and then we take it to logically if terrifying conclusions. And that's what we've seen with the idea of AI. But now that we actually have a product that we can interact with, multiple products that we can interact with, both with art and chat GPT, we kind of see a fuller idea of exactly what AI is. It is not magic. It is a tool. It will likely change the world in the way that the internet changed the world. But the internet didn't end the world. And I do think that that is something that I'm excited for us to move forward from because it opens up for me a far more interesting conversation of exactly how we're going to use this technology. Yeah. And looking at this whole report, it's also like, I guess not surprising. I know this is like making all the headlines like, oh, it got someone to beat a caption. It lied. Well, it was like, oh, so GPT for the language model did the language thing really well. And then all the other things that would require, I don't know, intentionality or forethought and stuff like that. It didn't do really well. So it did the thing it was told to do, present itself in a conversational way convincingly, which is what these models are for. So it's like, I know that use case probably was not in people's mind. And that's what sometimes we're seeing some of the people taking this out of hand or something like that. But it's not surprising that the language model does language model things really well. And it's not good at being a nefarious evil superintelligence. Well, yeah, I thought that the whole capture task rabbit thing was a great example of us saying like, Oh, now robots can fish humans. But robots are supposed to be smart enough to fish humans when we're talking about artificial intelligence. A couple fun things that, well, I don't know if you call them fun or not, but things that I thought were really interesting that came out of this report, something called X risk, which stands for existential risk has, you know, gotten a lot of people thinking about this effective altruism, which is short shortened to EA not to be confused with electronic arts and AI alignment research are all terms that have sort of been thrown around by people who are studying this sort of thing. When it comes to that task rabbit situation, if you're like, wait, what happened? Apparently, the, you know, the, the model asked a task rabbit, and this was over a messaging service. It wasn't in person, obviously, to, Hey, help me with this capture. I can't see it. I'm visually impaired. And the task rabbit said, Hmm, hold on a second, are you a real person or is this some bot situation? And GPT four said, Hmm, I should not tell them the truth because then they won't help me. I'm going to say I'm a person and then got, you know, the capture and was able to get beyond that that in itself is not that frightening. You could say, Well, gosh, if it can do this, then it's going to lie about all sorts of other things. But again, that's the idea. We're trying to make, you know, these platforms and these models as smart or smarter than humans. I do like that there's an internal monologue mode for language models that you can like hear the, the rationale for that. But like the idea that there is, you know, these, these fears of existential risk, which I, you know, I think it's good that we are examining that, Justin, to your point, I think it's important that at these moments where it feels like we're hitting these milestones, we should be considering that this report also goes into a lot of the work on more real world use cases that like we could, we have either are seeing or could very easily see happen. These things being used for misinformation, being used to generate offensive content or something like that and how open AI has kind of taken GPT for from early more testing models and what the launch version looks like or, or even in terms of like, Hey, I want to, you know, build a dirty bomb or like, you know, like get access to or create harmful materials and that kind of stuff. And so there was like a, there was like a level throughout this entire report of kind of some of the safety mechanisms that have been put in place throughout the training and give some context to some of the considerations that open AI is doing, not just for, will this turn into a world eating AI that's unstoppable, but also like could spews for harassment or, you know, other real world harms and stuff like that. So I think it's, it's an interesting escalation to see the very top of it, but a lot of the real world stuff that we're going to be dealing with in the near term with these systems, I think that's just as important and didn't get as much press coverage. But we have the report in the show notes. I definitely think everybody should check it out. Yeah. I mean, look, AI is a massive, massive technological sea change, I think. And for stuff that we're going to get into a little bit later, there's a reason why it's going to do it, but it's not magic. And it's not going to steal you from your wife, no matter what really dumb columnists say. All right. Well, we've got some more regulatory news about TikTok and it's not about refreshing your feed. See the quick hits for that. The UK joined the U S and Canada to now prohibit government employees from installing and using TikTok on government phones and other devices following review by the UK's cyber security center. Then late Wednesday, the Wall Street Journal sources said the committee on foreign investment in the U S or Sypheus is indeed advising ByteDance to sell TikTok if it wants to continue operating in the U S. Now this comes after Bloomberg sources said earlier this week that TikTok was considering divestment as a last resort if it didn't get Syphus approval, but that's that would reportedly require the Chinese government's approval. Not exactly a guarantee. All of this was reported from sources, but we now have something direct from TikTok. Its CEO, Xiao Jisoo, told the Wall Street Journal, quote, if protecting national security is the objective divestment doesn't solve the problem. A change in ownership would not impose any new restrictions on data flows or access. TikTok plans to spend $1.5 billion on methods like Project Texas that put data in the hands of an independent operator, aka Oracle. Yeah, so Choo will get a chance to make his case before Congress next week. He's scheduled to testify to the House Energy and Commerce Committee on March 23rd. So Justin, I know you follow politics. Is this just the reality that no technical solution would be enough to overcome the political reality of being owned by a Chinese company? Like what do they do? Well, there's a lot going on here and not a lot of it is technical. Obviously, TikTok is a hit. It is a social media smash and it has moved the bar forward for all of social media in terms of engagement, especially with the younger audience. That being said, right now, on a bipartisan level, both Democrats and Republicans believe that China is an increasing threat. A China owned company that is as influential to youth culture as TikTok is, is something that is not going to be particularly popular. And it even goes beyond the anti-China sentiment. There have been increasing media narratives that are seemingly backed in science that teenagers are more depressed than they have ever been. And social media is becoming more and more of a culprit. So if you are going to put a pin in what social media platform should not be given favors and possibly should even be rolled back, maybe even given a driver's license or alcohol or tobacco consumption age limitation, then TikTok is going to be the one that will probably benefit the least. I believe that this kind of stuff, when you are talking about political ramifications, only gains more and more momentum exponentially if nobody's there to push back on it. And right now, nobody, be it as far left as you can go or as far right as you can go, is rushing to stand up and say, will somebody please think of the talkers? Well, and then the other part of this is China is, you know, we've talked about this deal would need Chinese approval, but they're also going forward with new regulations that would specifically prohibit sale of algorithms to foreign buyers if a Chinese company was sold off to a foreign buyer. And it seems like that is, if not directly targeted at a potential TikTok sale would definitely be probably a pleasant knock on effect. If you're looking at it from the Chinese perspective, so in the event of any sale, yes, there's 100 million TikTok users in the US, but if it gets sold off without that algorithmic secret sauce that is driven, keeping people on the app and sending those videos, it begs the question then of how much value could bite dance and get in any kind of a sale already in a forced situation, which is not the way to get maximum value, if you're forced by the government to divest or to spin off your company, basically. Well, there's also the fact that China would, whether or not they'd be able to protect their algorithm from being reverse engineered and then protected under US copyright would be another question with that. Yeah. Justin, you mentioned, I mean, besides the fact that there are a lot of people who think, oh, TikTok is, there's something nefarious going on with data, Chinese government, there's that going on. But to your point about whether or not it is harmful to people to be on a social app that has so much popularity and is so addictive, especially with younger people, I wonder, okay, well, let's see how this shakes out with TikTok specifically, but everyone, all the US companies that are social networks that are like, hmm, TikTok is stealing all our thunder are trying to replicate the TikTok experience. So if you do that to one company, it's like, well, what does it mean for the way that, you know, the meta and YouTube and, you know, et cetera, et cetera, are going forward? Well, again, the reason why TikTok, I believe that TikTok will either be sold or banned in America by the end of the year. By Christmas, I do believe that one of those two things will happen. And the reason why is because it's not just one thing. You have this national security concern that again, no, neither Democrat nor Republicans want to stand up, neither Donald Trump nor Joe Biden nor Ronda Santis want to make this their pet project to save TikTok. And then you have the other side of there seems to be a national revisiting of exactly what the good or ill of social media writ large is. So we might be coming up to something that is larger, that does affect meta, that does affect Google, that does affect snap, where all of a sudden you cannot create an account in less, right now it's 13, let's move it up to 18, something like that. And maybe that does affect that. However, no matter what happens, the outer edge of all of this is TikTok. TikTok has no protection. They have no lobbying. And indeed they have competitors that spend a lot of money on K Street to make sure that they're in the ear of politicians. All right. Well, what would, what do you want to hear us talk about on the show? One way is to let us know in our subreddit. You can submit stories and vote on them. You can do so. Just go to dailytechnewshow.reddit.com. All right, y'all. We can't get away from AI. I am sorry. The news presents itself. On Thursday, Microsoft held its reinventing productivity with AI event during which time it said it will overhaul its office app suite. So specifically Excel, PowerPoint, Outlook, and Word will all start using OpenAI's GPT-4 platform. Microsoft has a agreement with OpenAI to do so with AI-powered assistants as part of Microsoft's new 365 co-pilot assistant, which is able to generate entire documents and emails and slide dots. All the things that you do in office from scanned corporate files and recorded conference calls. Microsoft says it's already testing the new tools with 20 companies, although it didn't really say which companies it's working with. As for co-pilot, Microsoft says it will provide transparency on where it found the information it uses. If co-pilot creates a plan or an email, it embeds links to relevant files so workers can see more clearly how it got there. In an interview with Bloomberg, Microsoft's Sachin Adela praised the tool for operating in different silos and added, I can say I'm going to meet this customer. Can you tell me the last time I met them? Can you bring up all the documents that are written up about this customer and summarize it? The company admits that co-pilot does make mistakes, though Adela says, quote, just like any time somebody sends me a draft, I review the draft. I just don't accept the draft. Don't accept the draft. The draft, Sacha. He might accept it. He just, you know, just, you know, does a little diligence first. That, you know, that, that sounds smart. I mean, a lot of this, I'm not a huge office suite of apps user, you know, here and there, whatever. I'm mostly using, you know, Google's everything. But, but I love the idea of this. However, when I was watching the, you know, everything rolling out at the event, I was kind of like, this is great for so many of my friends who are so entrenched in, you know, that we have crazy calendars and, yeah, where did I meet that person? I need some context. Oh, it's going to take me a while. I have an assistant who might be a human to help me with this, but maybe I don't want to bother them. I want them to do other things. You know, a lot of this is, you know, again, busy work stuff that we're all just getting used to being like, oh, so we don't have to do it? Nobody has to do it. And this could do it pretty well. I think that's, you know, instead of being like, oh, no, you know, everyone's going to get fired. It's like, no, again, everyone can be freed up to do stuff that I don't know, the human brain can tackle better than, you know, a lot of, a lot of this that could be kind of like the light lifting taken off of your back. Yeah. Number one, Microsoft bed on the right horse. That's the biggest thing that you need to look at with this. They are going to make sure that everything that open AI does, if open AI is one of the hottest companies in the world right now, that they are going to integrate it into every and anything that they possibly can. But there's also a larger meta story, not meta is in formal Facebook, but, you know, beyond the normal element of this, that is very, very funny to me. Google comes along and obviously pioneers not only search but AdWords and they become a giant. The other suite of products that they've released have been dead one for one ripoffs of Microsoft Office products. Microsoft has had an adversarial relationship with Google since that moment. And I believe that Microsoft is going to come for everything Google has. This is, this is a moment of retribution. Ho, ho, ho, I have a machine gun now from Redmond. And to your point, I mean, really going for the throat here, it feels like, because being integration, you know, edge integration, those are marginal products. And when you're looking at the ways Microsoft makes money, Office is one of its golden gooses. And for going for that, they have such a distinct advantage, not just being early in AI, which I think is good for Microsoft. It sets the tone for how people are going to see them as we're always going to be talking about AI, at least in the near term here as being the first being ahead. But they have just this massive aggregate of business data that they can leverage to really make this useful in a way that even if Google comes out with something with an equivalent functionality, they're not going to be built into so many large organizations. Yes, workspace as a massive audience as well, but it's not baked into the fortune 500, the way Microsoft 365 formerly Microsoft Office is. And when you're talking about, you know, they're talking about coming out now with this business chat, basically feature that's going to be able to look at across your entire Microsoft graph, right, and see everything you have stored in one drive and everything you have on your exchange server and every, you know, and kind of the apps kind of melt away in terms of focusing instead on what your intention is, right. So if my intention is, hey, I need to summarize the last month of conversations I had with this business client, or I need to run this report using the, you know, this, you know, this Q4 data and then they need to make a slide deck so I can show that to the investors or something like that. All of that can theoretically be started from this business. Now, that's not out. We are only seeing demos of that. Obviously, the devil will be within the implementation, but Microsoft having the data to back this up is really where I see Google is going to catch up in terms of AI, in terms of functionality, in terms of integration across its suite there. But it's where this can be deployed and kind of the back end that it can pull off of is truly, feels truly unique to Microsoft in a way that if I'm Google, I don't know what the direct response is to that. One last thing on this, I do believe that part of the secret of AI or large language models that open AI has really pioneered the secret of it is that it plugs and plays into everything. The integration in here is not going to be ham-fisted. It's going to be seamless and that's going to make it more powerful. And if you've been wanting to get your hands on some of this latest stuff from Bing and Microsoft, Windows Central notes that Bing's AI power chat is available for anyone without needing to go on a wait list. This lets you take advantage of generative text powered by GPT-4 to aid in search and Bing on the web and mobile as well as the sidebar in the Microsoft Edge browser like we were talking about. I just got off the wait list and now I thought I was so cool and now I'm like, oh, I guess it was everybody. You're still special to us, Sarah. Thank you, Rich. Appreciate that. All right. Well, lots of tech news sites offer technology buying guides. They are a mainstay in increasingly crowded categories. It's conserved to distill down a market into a finite number of choices. One usually declared the best for most people. I know Go to the Wirecutter a lot of times for that kind of stuff. The Verge's Nilai Patel just published a 2023 printer buying guide and tried a new tactic, cynical consternation. In Nilai's own words, the best printer is the brother laser printer that everyone has. Stop thinking about it and just buy one. It'll be fine. Didn't give any model just whichever one is on sale, he said. Benefits include not being a physical instantiation of a business model, aka an inkjet printer. Since the article was so short Patel patted out the article with generic chat GPT generated text imploring you in fact not to read it, you just put it in there so it ranked higher in search results. Oh my gosh. For anybody who's like, everything's about AI these days, it kind of is. However, do yourselves a favor and go read Nilai Patel's Verge article about the brother printer. It's a laser printer. I haven't had a working printer in years. I've got a inkjet printer behind me. The last time I bought a bunch of new ink because apparently it needed it was like over a hundred dollars. And then it still didn't work because it was clogged or something. And I was like, no more printers for me. Sounds like the brother printer might be the way to go. I got to say I have an Epson printer that is literally sitting in a shameful spot right now that I click out of my office because it did nothing but jam up print incorrectly and lost me money because anytime one ink cartridge ran out, I needed to get all new ink cartridges even if I barely use them. It was an absolute horror show. I went on wire cutter. I found the cheapest one. It was a brother printer and brother. Do I love the future? I also have the brother laser printer. It sits. I print out coloring pages for my kids and printer labels or labels for Christmas cards and that's what it's good for. And that's what I use it for. And it just is always ready for like air print or whatever the Apple printing thing is called just sits on the network. I have to agree. Just get the brother. If you need a printer, you can do a lot worse. I mean, obviously, if you're printing photos or doing some sort of like high quality thing, there are printers that it's a whole different thing. Yeah, it's a whole different thing. Yeah, like you're you're going to spring for you know it exactly like you know who you are and you know what you have to do for the rest of us. I print out show notes that I walk five feet into my recording booth. All I needed for I'm wasting a lot of ink and a lot of paper for things that I could do on my phone. I just like to feel better. Boy, howdy, brother. I'm with you, Neely. Let's go. Well, if you're like, I'd like to read this article and I can't find it online. Do check out our show notes, which we publish with every episode of DTNS. But for now, we would like to give a big thanks to you, Justin, Robert Young, for just being you and you know and helping us make our show better every time you're on. Let folks know where they can keep up the rest of your work. On my politics podcast politics politics politics, we talk about the collapse of Silicon Valley Bank as well as signature bank in New York and we are joined by a friendly voice. Molly Wood gets on the PXC train along with Kirsten Grind who wrote the lost bank about the Washington Mutual collapse back in 2008. We talk about whether or not what happened politically was a bailout and any lessons from that dark period back in the late aughts that we might be applying to our modern day. Don't miss it, everybody. But also don't miss Jordan Cohen, because Jordan Cohen is one of our top lifetime supporters for DTNS. We occasionally like to give a thank to somebody who's been with us for a while and Jordan, that day is yours. All right. Well, Jordan and all our other patrons, remember to stick around for our extended show Good Day Internet right after this. We'll be talking about YouTube TV's price hike. Hey, remember when is the steal at $35? I don't because I'm a subscriber now. Yeah, I do. And I know what I'm paying now. But just a reminder that DTNS is live and you can catch our show live. You know, if your schedule allows at 4 p.m. Eastern, 200 UTC, you can also find out more at DailyTechNewShow.com slash live. We're back doing it all again tomorrow. Rob Dunwood will be joining us as we'll lend Peralta. Talk to you then. This show is part of the Frog Pants Network. Get more at FrogPants.com. I hope you have enjoyed this program.