67: AI, AI, and more AI

[00:00:08] Speaker A: Welcome to Examining, a technology focused podcast that dives deep. I'm Eric Christiansen. [00:00:16] Speaker B: And I'm Chris Hans. [00:00:21] Speaker A: All right, welcome to another episode of the Examining podcast. I am Erik Christensen and I'm here with my colleague Chris Hands. Good afternoon, Chris. How are you this afternoon? [00:00:33] Speaker B: Doing well. This will be our first episode of the new year, other than the recap. [00:00:39] Speaker A: That's right. I forgot about that. I don't think of the years as different. They all blend together for me, but technically it's true. Yes. So we did the recap that we released, and then this will be the first, first one that comes out in 2025. And like many of the episodes over the last year, though, we did kind of drop the. No, I wouldn't say drop the ball intentionally paused. I'm going to use that. That sounds better. It's very AI heavy. So if you're not interested in AI, you're not going to like this episode, just turn it off. Because we don't have anything other than AI to talk about. And in fairness, there really isn't a whole lot in my mind to talk about that's not like tech rumors, which we typically don't do. So did you want to kick it off with one of yours? [00:01:29] Speaker B: Well, yeah, we've been kind of just watching the news cycle and everything is kind of AI focused. And one of the articles that I came across earlier in January was this one from Inc. Magazine, where AI isn't magic and it isn't a strategy either. And, you know, I think even if you just go based off of the. The title, without digging into it too deep, it seems like everybody's just going AI focused for any organization and the end of it. Yeah, it isn't a silver bullet. It isn't, you know, something that's just going to solve all your problems, but certainly it is a tool. I feel like, you know, technology is something that, you know, even prior to AI, you would have had to think about and find a way to become a little bit more efficient and so on. And again, even from a strategy standpoint, if everybody out there and all your competitors, if they're all using AI tools, you know, that's just now become the status quo. And so you, you still gotta go and figure out your, your, you know, root kind of issue and what you're trying to achieve and so on. I mean, I, I'll tell you, Eric, like, I even. I've met with a friend of mine recently and it was interesting. Like, he doesn't know much about the AI, like, we had lunch and stuff. But he wanted, if, if he had the power to do so. He's like, I want to get rid of my marketing team and my HR team and have like one person using AI tools to solve it. And again, I don't know if that's the best route either. [00:03:19] Speaker A: Well, okay, so there's a couple of things when I read this article I thought about, because it talks about and I should get the author's name because I don't want to not misquote it. So this was written by Larry Robertson. Larry Robertson. No, I've read some of Larry's articles before. So yeah, like you said, it says that in business, AI should not be mistaken for a standalone strategy. Kind of argues that a lot of businesses look at AI maybe because they don't fully understand it. They kind of overestimate what AI can do because I think it's not really intelligent. Because when I was younger, the way we described and defined AI was a little bit different. Right. These, the idea of something that could reason doesn't really do that. Generative AI, it's more of a text probability compiler. Right. So I think he says it shouldn't work as a cure all. It's also interesting, I kind of reading between the lines, it also sounds like something that Paul Thurot, the Windows reporter who we'll come back to, talked about because he had written that piece that we covered some time ago about how he won't pay for AI and he's like, you know, if a price on something goes up and because they're integrating all these new features, including AI, well then that's fine, but it should be part of the cost of that product. The AI is not the product. It's a feature of when it's integrated into something else. And I think that's also part of the problem. The fundamental economics of a business don't really change just because we've integrated AI. So if you, if the problem in a company is some pain, has, is some economic issue, you know, injecting a generative AI into some department isn't going to fix anything. [00:05:01] Speaker B: Yeah, yeah, exactly. [00:05:04] Speaker A: And so I wonder if that's what it is. And to your point about, yeah, cutting teams, I also wonder about like liability. So if you have a marketing team, let's say you get taken to court, you make a product for misleading the customer. I mean, these things happen. What do you, what do you do? Like if you offload this stuff to AI, okay. So you'll save money on, you know, key people. I Suppose in the short run, but like, what is the liability? Do you just say, well, I'm not responsible because, you know, go talk to OpenAI's model. I don't think it works like that. [00:05:42] Speaker B: No, no it doesn't. I mean ultimately you're, you're accountable and liable for anything that you do. You can't go and blame it on a, an algorithm. It's the same thing if, look at these self driving cars. I mean, if you got into an accident or were speeding or whatever, are you going to blame Tesla's, you know, self driving mode? Full self driving mode. I mean at the end of the day you're, you're the one who owns the, the vehicle and you're the one who has the driver's license and are responsible. Right. [00:06:13] Speaker A: So, but I just, I just think of things like marketing or especially hr. I mean it's, it can be pretty sensitive. Do you really trust. I mean again, we're not anti AI. You and I have been experimenting with it. We're trying to figure out where it's really valuable, where it's not. We're cautiously enthusiastic, I suppose, but I, you know, there's also real professionals who work in marketing and AI who know what they're doing. And so to offload it, you're going to get perhaps mediocre and perhaps really generic results. That would be another thing I'd be afraid of. Like, if you're using the. Like, what about what's so special about one company's AI to the other? I mean, it's hard for me to imagine that they're not all pretty much using the same five, one of the same five models. [00:06:56] Speaker B: Well, and I mean, even so what. [00:06:58] Speaker A: Value do you bring if everyone's using, you know, it's like, it's almost like if you all used OpenAI, all these companies competing in whatever industry you all hired, that you've all outsourced it to the same bot basically, so isn't going to become generic. [00:07:13] Speaker B: And that's where I think you even mentioned like offline to me. Like, you know, you logged into LinkedIn after a long time and a lot of the posts. [00:07:21] Speaker A: No, we could talk about this. [00:07:23] Speaker B: A lot of the posts seem very similar. Right. And even like the, I don't know again, they're probably taking like some crappy prompts and putting them together. I mean, personally, again, I've looked at it. I mean, some people have told me that maybe I should still keep posting regularly because you're building your, your brand and all this kind of Stuff, but. [00:07:45] Speaker A: I, I don't know if I believe that anymore. I mean, I hate to, I don't mean to be interruptive because I remember you and I both heard Scott Galloway talk about that. You know, pick the pla. Pick one of the platforms that you're best suited to creating content for. If you're good at video, do YouTube, if you're good at, I don't know, motivational posts, go to LinkedIn. I don't know what LinkedIn is right now. And then, you know, try to be one of the top, you know, 10% content creators. But that doesn't scale for everybody. And to Cal Newport's point about productivity, he's actually on his podcast, people have said, you know, do I need to be on LinkedIn? I mean, I have it, I keep contacts, like a list of contacts there. It's a great, like, yeah, contacts app. It's basically what it's for. And you can add people and see their cvs. It's great for recruiting and jobs. I think it has a lot of value. But do people have to be there to be successful? Is it, is the implication, like, if you're not creating lots of content on a platform, you're never going to get work again, or you're, you know, I don't know, I have a hard time believing this. I think that like blogs, for instance, are built probably by more like sharing and word of mouth more than the noise of some feed. [00:08:57] Speaker B: Well, exactly. And I mean, at the end of it, look at like Cal Newport's a good example. Like he's not on any of the platforms, right? [00:09:04] Speaker A: Yeah. And he's a bestselling author. [00:09:06] Speaker B: Exactly. So I mean, he's a best selling author, he's producing. I mean, his podcast is generating revenue and so on. So, and, and this is where I, I feel again, if, if somebody has the time, you could do it. I mean, look at, we could be posting all day long just like how all these people are just probably running it through some AI writing tool and just posting stuff all the time and they all look the same. Or we can just, instead of, Even if it's like 15 minutes every day, like I'd rather take that 15 minutes and maybe spend it with my family or, you know, take care of some tasks or something else. Right. Like, so again, that's kind of where I've shifted because I, I'm busy with other things that are taking more priority than, you know, social media. And again, I'm not convinced. Maybe I'm in a different stage of my Career. I don't know how much real benefit that it's going to give me by going and you know, making that one post a day or what have you. I mean, I don't think it would hurt, but it takes time, right? [00:10:12] Speaker A: Yeah, I mean like I, if I blogged more regularly and my blog posts that I do post, I mean my Techbytes blog that you as an example. I very rarely shared any of it on social, but I got like thousands and thousands of unique views every month. 10,000. Yeah, and I guess I had it for a long time and it was indexed on Google and people probably bookmarked it or added it to their RSS feed or, or they, or I have it so people can sign up and it's like, hey, do you want put in your email? I don't collect any of the information, I don't care, I have a list. But you know, it'll just email you when I posted something new. [00:10:48] Speaker B: Yeah, totally. [00:10:49] Speaker A: In fact, I think that's better. I mean I don't follow any blog. I get notifications for Medium because I follow like UX Collective and a few things like that but. And I pay for the, not the top tier but the kind of middle of the road so that people get some money for what they post. I do republish things to Medium occasionally because it's a format people like. But I don't really do a lot of social advertising, just our podcast really because some people follow it there which is fine. [00:11:16] Speaker B: Yeah. [00:11:18] Speaker A: And now it's easy because I just have AI and make those posts and maybe I'm part of the problem. [00:11:24] Speaker B: Well, and then you know, beyond that, like I again this social media, like this, this whole social networking aspect hasn't been around that long. I mean, and then I guess some, we didn't include this as part of the articles but I mean I also find it funny. TikTok got banned, then it's unbanned, I guess for like 75 days. I saw an article today actually that some people were selling phones with TikTok on there, like just so you, you know, downloaded and they're selling over thousands of dollars like their, let's say their iPhone 15 Pro or whatever with TikTok already installed because apparently there's people that are, I don't know, content creators that if they deleted it now they might need the app if it continues. And this is only a 75 day kind of thing. But I mean in some respects we've talked about this in the past but like I think from a security national security perspective, it probably might be a good thing. I think the other thing is just from a societal standpoint, I've even talked to students these days and they're not even watching TV or anything. What they're doing is they're just on TikTok. That's their form of video entertainment. [00:12:43] Speaker A: So we will get into other articles. I know I didn't want to do too much of a tangent, but I'm listening to what you're talking about and I can't get out of my mind just how much content is created on TikTok and LinkedIn and I guess Instagram, I don't have a lot of those anymore. I don't really. I haven't looked at the Twitter feed forever. I know that there's some good stuff out there and there's people who do really good work and share things. So I'm not at all disparaging. But have you ever thought, and perhaps this is just me. So if it is, just tell me that a huge percentage, whatever percent it may be of whatever comes through on social networks seems just, just like a huge grift. I see, I see so much. It's just inundated with like self help, you know, productivity hacks that aren't based in any fact. I mean, we talk about productivity, but you know, we typically cite books, we talk about studies that have been done, or we bring in our own experience of best practices. You and I are, I think, do a pretty good job of sourcing things outside of the tangents and conversations that I have, but just so much lifestyle content that doesn't seem to have a lot of substance. And I find that most of the feeds I have on most of the networks are that. [00:14:20] Speaker B: Yeah, no, I, I would agree with you. And that's why personally, like for, for instance, like most of the apps, I've actually like all the meta apps, I don't have them on my phone and so I rarely log in. [00:14:31] Speaker A: That's good. [00:14:33] Speaker B: You know, even now on my phone, what I've done is I've created a folder on my third page of my apps with and the folder social networking. And then I put those like let's say Twitter and LinkedIn in there. I turned off all the notifications. And I'll tell you, just by virtue of putting it in that third page, putting it in that folder, I don't check it. And so only if there's something that I need to go and check, that's the only reason why I've left it on there. If it's something on the go or something quick. But I mean it saved me a huge amount of time and it's again, I think this is where the, the social networking is a huge time vampire in society. [00:15:14] Speaker A: Yeah. I would rather take that time into doing more episodes with you and citing them and releasing them like we're actually making. I'm not saying that we're the best of all the podcasts. I mean we aspire to be. But there's people with more experience, better video editing skills. But at least we're making in depth content. Like long. What I mean is long form content. Content that's not meant to be scrolled through. Right. I guess that's what I'm interested. I'm looking for depth. [00:15:40] Speaker B: Yeah, absolutely. [00:15:43] Speaker A: But that's a good. Perhaps a good. We can delve into that more at another time. Did you want to talk about the Ars Technica or did you want to move to Reuters? [00:15:53] Speaker B: I mean we can just kind of quickly talk about it. So there was this article that just was talking about how medical information misinformation could be done with large language models. And I think if you look at it again it's, it goes back to. And this is where I think it's going to be a nice tie into one of our later articles. But it's kind of like that old adage even some of my students have been telling me this and just people in industry that they're finding the results and the outputs that are coming out content wise out of these large language models isn't that great? And so, so and I, I think it's almost like that. Remember when we did this talk last. Last year at the faculty retreat and it's like you know, making a photocopy of a photocopy of a photocopy and then you. I mean that's kind of like the analogy but at some point this AI generated content is just creating more garbage. And so this particular article was talking about a study where they looked at a medical research and how you could. It could actually be a good place for misinformation. [00:17:06] Speaker A: Well, and how easy it is to derail because didn't they say in this piece it was like it only requires.001% of the information like the training data to be misinformation to totally throw the thing off. [00:17:21] Speaker B: Well, and again it's, it's like it. [00:17:24] Speaker A: Seems like it would be way more than that percentage. [00:17:26] Speaker B: But I think that's where like this again the transformer architecture that's created there because this and this is where I, I don't, I don't know why even just in society we should be using more critical analysis, critical thinking skills. But like your point earlier, this AI and that maybe the definitions have changed, right. But this generative AI is not real knowledge. It isn't, you know, it's, it's literally figuring out probabilities of words that go together. And so based on what you put in, then people are just taking it as yet. This is, it knows, you know, this. [00:18:03] Speaker A: Is, yeah, it's good. If you limit the corpus of knowledge, like scan my CV and pull out leadership experience over the past 5 years or something like that. Like it's really good at that because it's, you've given it, it's very, it's very clear the boundaries of the information and you've given it very clear instructions. I mean to me that's where the AI stuff is the most interesting. It actually has nothing to do with. There's a lot of magic out there. It's a little bit too much magic discussions with AI. But when you give it clear data to work with, you know, organize things like this into categories. Yeah, you have to, you can edit things, but it does a remarkably good job at stuff like that. And that's to me better than kind of these huge data sets and these kind of almost, you know what I mean? Like these hugely aspirational goals of what LLMs can do. To me, an LLM that ran locally, that just analyzed documents that I inputted into it would be perfect. [00:19:01] Speaker B: Or like how we've been talking about how I've been, you know, telling and you know, encouraging my students. I think from a business communication standpoint, it's, it's a great tool. You know, that's. Oh yeah, that's where I think right now, where the technology is that that's probably the most relevant in the workplace is just taking care of your day to day stuff to cut down and make you a little bit more efficient so you can actually spend time on the things that are important and of value to the organization. [00:19:33] Speaker A: So yeah, like maybe you're, maybe you're a solopreneur and you have to put out press releases or something or that's not your skill set and you're trying to change the tone. It's like, well this is a lot more efficient or something. Like, I mean these are, it's not necessarily even taking away a job. It's because you might never have been able to hire somebody to do that. I'm never going to hire the Images are more contentious and I respect and I understand that perspective. I'm never going to hire an illustrator to make images for my blog posts. It's never going to happen. So I often generate them and I've generated header images for my website for presentations. I generated a watercolor image of someone standing giving a lecture because it makes it more interesting. Sometimes I use open licensed images, but sometimes I can't find what I want, so I have to generate it. Right? [00:20:21] Speaker B: Yeah, exactly. [00:20:22] Speaker A: And you know, I never would have hired someone to do that. It just, it just gives more flair to something that I would have had to otherwise do myself, maybe badly. Right. So yeah, there's a. Speaking of magic, so there's this huge investment. Have you heard about this, this big $500 billion announcement for this Stargate project? This, this big AI investment. By the way, why do these AI investments have to have, like, why Stargate? Is this supposed to be like the TV show? Did they choose that because they didn't want to say Skynet, which is what it really is. Like, I find this to be. I think there's been so many science fiction movies made about AI, they should be using the AI to name these projects because whatever they're coming up with is no good. [00:21:16] Speaker B: Yeah, exactly. I don't know if we mentioned this before, but like even with OpenAI, so with chat, GPT now they're probably gonna just drop the GPT because they did they. [00:21:33] Speaker A: So what's it going to be? [00:21:34] Speaker B: They're just going to refer to it as chat. And so. [00:21:37] Speaker A: But chat like OpenAI chat or AI chat, like they're gonna have to brand it. [00:21:43] Speaker B: I don't know, I think they're just calling a chat right now even it's. And this is where like again I find personally, like when I'm talking to young people, like that's what they're referring to it as, right? So they're, they're just calling it chat. But the domain was owned by one of the co founders of HubSpot. And so in exchange for the domain, he apparently, according to some of the coverage, got a bunch of shares in exchange for that. So yeah, it's kind of interesting. But yeah, I mean, I don't know, we've talked about from a branding perspective, like, yeah, GPT is not the best kind of thing and then. [00:22:23] Speaker A: But the best brand is the one everyone remembers. I mean, Coca Cola doesn't make any sense to me either. But like it's stuck. They can't change it. Now to me, GPT it is what it is. Even if nobody knows what it means. It's. Yeah, that's the name. [00:22:40] Speaker B: Although I think the. [00:22:41] Speaker A: I think people are gonna call. It's like Twitter. You can't change the name. It's too late. Yeah, nobody. Nobody I know calls it X. I don't know anybody who says that. [00:22:49] Speaker B: I agree. But yeah, I mean, I think Copilot was probably the best and more intuitive name. But I guess in terms of this. This investment. Yeah. I mean, I don't know. I like the Stargate show, but I don't think this has anything to do with star. But that's too bad. [00:23:08] Speaker A: I thought maybe you materialized in from one data center to another. I guess we should explain what it is. So it's a bunch of $500 billion to build AI data centers. And then the first phase is 100 billion, so a fifth to create them in Texas, and then that's supposed to expand across the United States over how many years was it? Four years. Seems extremely ambitious. But maybe it's easy to put together a data set in it now. I don't know. What do I know? And it was supposed to be some of the applications were, you know, of course, AI applications. So that would be the chat GPTs, that stuff, the chat bots. But also like maintaining electronic health records, assisting medical professionals in collaboration, detecting cancer, personalized vaccines. This is where my eyes started to glaze over because, well, okay, how that seems like an application. AI application to me. All. Everything's an AI application. [00:24:10] Speaker B: Right. [00:24:11] Speaker A: So I don't really know how what we have now couldn't do that and these would allow that. But maybe at scale, who knows? [00:24:18] Speaker B: I mean, when Larry Ellison was talking about that, I mean, it just seemed like some snake oil kind of stuff. But yeah, I mean, I, I think from the government perspective, probably the. Beyond the healthcare and the economic growth problem, probably for the US it's the national security aspects and especially with cyber security and military. Military and defense kind of mechanisms. So. But yeah, I mean, it's. There has been as of late, I saw Elon Musk was kind of against this whole. He's expressed a lot of skepticism regarding the 500 billion and especially the, you know, how the government is supporting just a small group. I mean, it almost. It seems like, you know, back in the day where we had like the, the oil barons or the, the railways, it's like. [00:25:12] Speaker A: Yeah. I mean, there was contracts given out. Yeah. I guess that all depends on like how they, how they determine who's part of the partnership. Right. I mean, that's one of the problems in government. It's not supposed to be. I mean, well, it is, but it's not supposed to be patronage. It's supposed to be, you know, that's why we have requests for proposals and stuff like that. I don't know how this materialized. It just seemed to come out of nowhere. It's like they said, anybody got $500 billion to throw at this for a news coverage? [00:25:39] Speaker B: Well, I mean, again, I mean whatever people think about Trump or what have you, but this, I think the big, and I can't remember I read, I came across some like news coverage on this. But one of the reasons why the US has been very dominant worldwide from a business perspective is they're willing to take risks. They're willing to take like big, big bets on things. And you know, you certainly outsourcing things to like China and stuff. I mean they literally have access to all our technology. But still somehow we're, you know, the Western world is still doing pretty well. You know, you look at like from, let's say from smartphones or other devices and other applications and really the big thing is this AI focus and this should beyond all their like self interest, but it might create more economic wealth for the, the country and at least strategically, maybe they'll have an edge over other countries. [00:26:47] Speaker A: Yeah, I mean historically the US has had an advantage because they've had a very large middle class to sell to. Right. So I think one of the concerns, the, the differences this time, without getting into the economics of it, is that it the early 20th century, up until, you know, my, I guess the start of my lifetime or well into it, the middle class in the United States was really the driver of the economy. I think one of the problems is, is that as you kind of get wealth disparity and you kind of hollow out that the return on investment becomes harder and harder because it's just fewer and fewer people to buy the product. Right. So it seems to me like if anything the Stargate project seems more like something that you would see out of China where it's like a government corporate partnership because they don't have that. It's not like a, it's not, it's not a consumer product that, you know what I mean? Like, this isn't like a, a growth driver from like an industry perspective. This is like a government partnership, public private partnership, which it could be good depending if they can afford it. I don't know if I'm really explaining this right. It just, it wasn't, it wasn't clear to me how this is funded or what it is. This is very few details, but it seemed more like a command economy project to me than a bunch of companies getting together and doing what they will. A shareholder money alone. Yeah, I should say. [00:28:17] Speaker B: Well, and that's part of the. What Elon Musk is saying is especially the, the, the companies involved, they're not putting much money into this. And so it's mostly public funds that are going into this. So. [00:28:29] Speaker A: And I mean in fairness that there's lots of examples of public works in the United States too. I mean the hist. The Hoover Dam. A lot of the great public works the United States are built during the Depression or well before that, especially during downturns funded by both parties. So it's not. Public works are valuable, especially if they have a public benefit. Electrification of rural America as a good example as a public works that really move things forward. But is this going to do that? I guess is my question right. Or is it really just for a handful of people in a handful of sectors? I don't know. It would be great if they built out data centers and then gave. Because it's public money and public investment businesses more free or very low cost access to use AI in those data centers. Especially if the data centers were powered by nuclear rather than like coal or something like that. Because ideally it's kind of like cities who do the green bin system. They allow people to get compost for either free or very low cost because of course they're providing the material and the labor. Right. So if it's a publicly driven project, I would, I would hope that they say okay, businesses get X many credit. AI credits or hours. Yeah, sounds like a timeshare. But something like that where everybody could make use of it because of course there's a lot of these technologies out of reach to a lot of small companies. Right. I don't know exactly if that's the idea. It wasn't clear to me how it's going to work. [00:29:55] Speaker B: Yeah. Who knows? But it got a lot of press coverage because this is a huge amount. 500 billion isn't something that you normally. [00:30:05] Speaker A: And we're talking about it so clearly it's getting like. Because all the best news comes here. So I mean like this is. We're on that. We're at the tip of the spear. So if you hear it from here, you know, it's, you know, it's good. We had some other sources to, to move into. So the mobile syrup had an article about this though it's available in other places. Essentially there's Chat GPT and this is only available in the United States right now. But there's a new feature called operator. So the OpenAI operator uses AI to look at web pages and interact with them by typing, clicking and scrolling. This is pretty funny actually in my mind when I first read this, I had a vision of the webpage up and then the guy was like, move the mouse x many pixels to the left and click the button. It's something incredibly tedious, but apparently it's supposed to be able to go and do things for you. So like, okay, I need this stuff from Amazon, go buy it and have it shipped to this address with free shipping, you know, prompting things on the other end. It says it's powered, so it uses the same GPT4O model. And the operator works by seeing kind of like these screenshots and then interacting with them by using actions on a mouse and keyboard. So it's like an interesting. But it's, it's for GPT Pro users, which you said is, is that in, that's only in the States, the Pro. It's, this is beyond. [00:31:40] Speaker B: And apparently it's, it's going to cost. So it's, it's in addition to what you're paying right now. So it's actually a 200 per month plan as opposed to your, your base version that you had for $20. And so again it allows you. And so this is the big new trend where they're referring to as AI agents. So these, the AI is acting on your behalf and doing stuff. I mean, I, it'll be interesting to see. I don't know if I would trust an algorithm, especially when we know that it's not, it's, it's basing it off of a data set or whatever. But would you trust this thing to go and find you and book like, I don't know, through TripAdvisors on trip and all the flights and everything like that and just let it go. I mean, you could give it a shot. [00:32:30] Speaker A: No, I can, I, I can picture myself wanting to book a vacation and I find myself on a flight to Yemen or something. I, I don't, I don't understand. I, I thought the AI agent. Well, I, I should, I should refrain from details because I don't. When I heard Microsoft talk about AI autonomous agents, it seemed more in the context of kind of enterprise business going out and finding things on your behalf. Kind of like it could do things like Google alerts, it could alert you when certain triggers Happen maybe you're working in network security. So kind of a co pilot that goes off and. Or I think of somebody who does security that goes patrols and then reports back. I didn't think it was actually going to buy stuff and have signing authority over people. I find that to be, I mean I guess it's. [00:33:24] Speaker B: And it is doable again it's. Who knows, people will probably do it. It's just like with the, the full self driving mode. You know people, there's many people, I've, I've seen them record videos. I've seen it personally. They go and do it. They have the car come to you, you know, as, as a valet, self valet. I mean that's not legal to do that but people do it and so I'm sure there'll be some early adopters, they'll give it a shot and see what happens and they'll probably be impressed. But yeah, I mean I think that's where what they're trying to pitch is this where you don't have to deal with any kind of menial tasks. [00:34:09] Speaker A: I find the idea of an autonomous agent that helps me with research helpful. Keep an eye on the latest articles published on this. This is a topic I'm interested in. Help me compile them into a database so I can search them later and save them as PDFs or something. I can see all sorts of tasks that I could have something do proactively and then I could go and review it at the same time though maybe this is just me. A world where there's no menial tasks for whatever reason to me seems somewhat unhealthy. I mean there's something good I think about having to, you know, struggle a little bit and to figure things out. I like doing a lot of menial tasks. I find that's when I get all my best ideas. You know my favorite research paper came as an idea to me when I was sweeping or cutting a lawn or something like that. I mean I just find those things to be helpful. So I don't know the vision for these AI agents. I can't say that I totally understand it. There is a longish video that's about 20 plus minutes that people can watch as part of this article and it kind of explains how the tool operates. There was an interesting article from during Fireball so Jon Gruber's blog. This one I found really interesting because it talks about how Siri has actually gotten worse as a result of the interaction or the, sorry the integration of the Apple intelligence, specifically the chat GPT features. So in this critique, it basically says that Siri has declining performance, particularly in providing accurate responses to straightforward queries. So he references kind of this experiment that was done by Paul. I don't want to get his name, his last name wrong, but cathesis. So he asked Seri about winners of the Super bowl one through. What is LX in Roman numerals? 40. This is something to look up, 60. But I don't think we've done 60 Super Bowls yet. So I think that's the kicker. So Siri correctly identified the winner in only 20 of the 58 instances. So that's a 34% accuracy rating. And then he said Siri erroneously credited the Philadelphia Eagles with 33 Super bowl victories they haven't achieved. And then he said he contrasts kind of these failings of Siri kind of with the proficiency of other AI systems. So he compared it to chat, GPT and other online search engines that are not Google. So Kaji duckduckgo. And they handled all of those queries with way more accuracy. So, yeah, it's interesting to me that Siri was more limited for a long time, but I found it to be typically right, like, when is this game going to play? Who won this? How old is this person? How tall is the celebrity? He was pretty good at that stuff. But apparently, however they've integrated it with ChatGPT and Apple Intelligence just is not working. And it's kind of sad because it's, I guess, what this article reminded me of because. And it kind of goes back to what we were talking about with the volume of content on social. Sometimes when I'm using these tools, like, especially if I'm not looking for a particular answer or I'm not asking it to analyze something, I'll just be asking it general questions. It gives me a lot of content, like just a lot of bs, but it's not very targeted or to the point. Right. And so I wonder if older AI systems, where perhaps you had to memorize more precise voice prompts, you know, Amazon's Alexa kind of works like that. Yeah, they were limited, but what they did was more consistently. Right. [00:38:42] Speaker B: Well. And you know, it's definitely a problem. I mean, I think with strategically for them, they did this partnership with OpenAI just so that they can go and, you know, not be left behind. And obviously we, we figured, and they said as much, that they're going to be working on developing their own GPT or some sort of version of technology for themselves. But I've noticed that myself is even just Basic things Eric. Like I try to go and search on my phone just through notes and it won't bring up the, the note and I know it's there and so I, I don't know. [00:39:24] Speaker A: So is that a feature of, of it, of, of Siri with the Apple Intelligence? [00:39:30] Speaker B: I don't know. I mean it's, it's only happened in the last like let's say month, month and a half where I literally, I know that there's a document or a note that I've created in Apple Notes and I can't even find it and then if I go on my computer it's a little bit better. So it's, I, it doesn't make much sense to me but definitely there's, there's issues with Siri, the whole AI team and just a few days ago now they've made a big executive change so they've put in place this Kim Verath who previously she was the one who took care of the Vision Pro software and so she's becoming the, you could call it the, the top deputy for the Apple Artificial intelligence machine learning division to go and get this figured out, this whole Siri stuff. So you know, it definitely is something top of mind for them and hopefully they'll be able to figure it out. [00:40:30] Speaker A: Yeah, I don't know if it's best to set it up with Chat GPT or not because then I guess your information is going somewhere. Have you done that? [00:40:38] Speaker B: I haven't, I haven't gone and integrated it or anything but I don't think I want to. I don't know if I would want to either. I mean even I've turned off some of the. After I noticed that it was having some issues even finding notes and stuff, I went into the settings and told it not to learn from my, my activities on my phone. [00:41:03] Speaker A: Can you actually do that? [00:41:04] Speaker B: Yeah, there's, there's options like there's different apps they use it can you can go and have it go and submit it to Apple Intelligence to go and learn your behaviors and get better. But I, I don't know if there's much, I mean I'm, I'm pretty, I try to like we, we chatted right before we started doing this. I mean I think there's, it's better just to go and find that one or two tools that you use and know it well and just focus in on that and instead of just having to constantly be learning all this stuff, just get your work done. Right. [00:41:41] Speaker A: I agree. Well, that's most of the Articles we had today, I think for news, I had a tip that I thought might be of use to people. Also AI related. Microsoft's 365. What used to be Office 365 is renamed Microsoft 365 a couple of years ago. You. It now gives the primary user, so not everybody in a family plan and in a personal plan, it gives the primary user Microsoft Copilot. So Copilot Pro was the ChatGPT. Well, not really competitor because it uses ChatGPT or the latest models. So that was again, similar price, that $25 a month or whatever. I haven't found it to be quite as good. But if you have Office 365, which I do, it now integrates that automatically. So a lot of the writing tools, the advanced features on the Copilot app, image creation, things like that. The difference is that you just get a certain number of Credits, which is 60. You can go to your Microsoft account and it'll actually tell you how many credits you've used. It's like one credit per prompt, essentially. So. But that actually is included. It's included as a larger price hike for Office 365. So it used to be like $100 a year for the family plan and now it's maybe a bit more. But then it adds that feature to everything. So every app, you can put Copilot in it. So Office, you can have it running side by side. But that's the tip I have. So if you're hesitant with this stuff, you're in OneNote or a tool or maybe Outlook and you want to help it compose and do replies and summaries. It's actually integrated up to a certain point. And then of course, after you've used up your credits, you have to have to wait a month or you have to pay for the Copilot Pro. [00:43:39] Speaker B: Cool. [00:43:40] Speaker A: But just kind of get better example of them integrating it into something and factoring in the cost rather than making it the product. [00:43:48] Speaker B: Yeah, makes sense. I mean, even Satya Nidadella of Microsoft, one of the things that he's now, and a lot of people are like paid attention to this, but he said the software as a service. So SaaS is done, it's over. And so you know you're gonna have to go and find some other business model just because of this technology. [00:44:09] Speaker A: And so why would software as a service be done? I don't get it. [00:44:13] Speaker B: Well, he thinks that. Let me just bring this off. [00:44:16] Speaker A: Isn't AI software. [00:44:18] Speaker B: But I think with all these other companies or not other companies. But with the, the AI subscription basically negates the need for any other subscriptions. I think that was his logic. Let me just bring it up here. [00:44:40] Speaker A: I don't know how it's going to negate other subscriptions that I have, but that's fine. There's all sorts of subscription software as a service. [00:44:51] Speaker B: And it can be perceived as a meta, metaphorical way of saying that it's high time for SaaS business model to evolve and emerge or adapt to emerging technologies. [00:45:04] Speaker A: Interesting. I will read that because I have not read that article. Well, that seems like pretty much everything we had to discuss today. [00:45:12] Speaker B: Well, there was one more tip that you had. You shared with me right before the, the web search, remember? [00:45:21] Speaker A: Oh, right. I can't. I can talk about that. Yeah. So I guess I'll include it as a tip. Mostly because you didn't know how to do it, which I'm kind of surprised because you're way ahead of me on this typically. But if you go into ChatGPT and I was listening to a podcast, so if you start a new chat, so I don't know if this is in plus or if it's in regular as well. Let's log out and try. But if you're in plus for sure, under the prompt there's a little globe icon, which is an Internet search icon. And so if you click that and then do your prompt, it'll actually search the web. So ChatGPT has some sort of index, like web indexing that they do that's, you know, they're equivalent to Google. And so you can go out and it'll do things for you. And it typically brings like pretty detailed responses back. I mean, you can kind of ignore the responses it gives if you like, but it will give a considerable number of sources. So in my testing, it's often given me like 20, 30 up to websites if there's a very contemporary topic. And so that's super great. So if you don't, you're not interested so much in its, its actual answer, but you're interested in the, the sources it gives. This is actually often a really good shortcut. I did find it worked a little bit better for older things because I'm assuming that their indexing isn't quite up to speed as like a search engine company. So I, I searched for some kind of older articles on older technologies and I found some terrific tech articles. Like I was looking for tips on Raspberry PI projects and it brought back a ton of stuff that I might not have found on my own, again, some erroneous sources, but it didn't filter them for me. It just. It just gave me a list. So it's kind of an interesting web search alternative. [00:47:22] Speaker B: Yeah, yeah. No, and. And personally, like, I just didn't notice, like, these icons at the bottom that you had to enable them. And I figured that because when I was searching for stuff, it was giving me sources. But. [00:47:34] Speaker A: Yeah, well, it will to an extent, but that. That is, like, specifically. Just do a search. [00:47:39] Speaker B: Yeah. [00:47:40] Speaker A: And some. [00:47:40] Speaker B: Exactly. That's where, like, I. I found. Yeah. Like, this definitely would take you up into level if. And I mean, this is one more kind of bite or nail in the coffin for Google if it's now searching and finding the most relevant and better kind of sources on something that you're looking up. So. But, yeah, it looks like there's. They've just changed some of the icons because the one right next to them, the globe icon, there's. They have a picture tool now as well. And so that's for using Dolly. And then you have this, like, canvas thing. So, yeah, I guess they keep rolling out new product features or what have you. So. [00:48:21] Speaker A: Yeah, it's cool. So it's kind of an interesting test. [00:48:24] Speaker B: Absolutely. [00:48:24] Speaker A: But that's pretty much all I had today. Where can people find you, Chris, to get in contact with you? [00:48:30] Speaker B: So if anybody wants to get a hold of me, you can find me on my website is Chris Hans. So K R I S H A N S CA My contact details are on there and various links and so on. [00:48:44] Speaker A: And it's the same story for me. So I'm Eric Christensen. Yeah. With a K. And it's eric christensen.net that has a link to all the other sites. I maintain this podcast, everything else. So with that, thanks very much for your time today, and we will ride again. [00:49:02] Speaker B: All right. It.