Episode Transcript
[00:00:08] Speaker A: Welcome to Examining, a technology focused podcast that dives deep.
I'm Eric Christensen.
[00:00:16] Speaker B: And I'm Chris Hans.
[00:00:24] Speaker A: And welcome to another episode of the examining podcast, the technology focused podcast that dives deep.
I'm Eric Christensen and I am back after several weeks with Chris Hans. How's it going, Chris?
[00:00:36] Speaker B: Good. It's nice to be back and you did a great job while. While I was away, so I enjoyed your short episodes.
[00:00:44] Speaker A: Oh, that's very kind. Yes. I tried to continue though, you know, it's usually every week or every other week, so the best that I can. I had to script it a little bit more closely and read that because it's hard to ad lib when you're talking to yourself, as I found out. So totally that's the secret to doing solo episod.
But I think we were going to cover three things. Today we're going to talk about your long term review of the iPhone Air since that's your new device.
We have a interesting time magazine person of the year and then a tip on running local LLMs.
[00:01:21] Speaker B: All right, sounds good. I guess I can start with the review on this phone.
So I decided, I mean, I had the iPhone 15 Pro and the price difference between the, the air and the new pros, it was only $200, but when I looked at it like the screen size, significantly bigger.
Right. And then also there was the weight difference. And so when I was thinking about it, I'm like, you know, at the end of the day, I mean, we talked about this previously when we were reviewing the new iPhone lineup.
I didn't think I would get the air, but when I was thinking about it, you know, my phone, the battery now it's. It was like 2 years old. I would have had to get it replaced. So there's like a hundred bucks or what have you that you to get Apple to replace it.
I thought let's just go and upgrade. And when I went into the store, I noticed like, yeah, the screen size is quite a bit bigger. I thought about some of the basic aspects of what I use the phone for. And personally, I don't particularly like the new design of the, the pros.
[00:02:35] Speaker A: It's a little chunky.
[00:02:36] Speaker B: It's kind of chunky. I think it looks kind of awkward with the bigger, you know, camera going.
[00:02:41] Speaker A: Right across the plateau. The plateau.
[00:02:44] Speaker B: And then, you know, even when I started thinking about it just in terms of my usage, I'm not a big photo person.
And so, you know, the, the pro lenses. Sure. Like, I mean, you can use it. I Mean, there's people who have used it even in. In movies and stuff, and that's. That's what it's meant for. It's meant for professionals. But for myself, here I am, I just need a device. And.
And so that was one of the reasons why I just decided to go with this as. As you get older, too, I think that screen size, it does help, I find, you know, especially going and reading certain stuff, like I have to go and hold it further away or what have you. And so just having a little bit extra screen space, that's kind of nice, I think, from a software standpoint.
Remember we talked about. There was that little kind of. I mean, I don't even know what they're calling it, the official name, but there's this little circle that brings up all the other, you know, like your control center and all that kind of stuff and notifications. I don't know if I find that really that much more helpful. It's kind of. Sometimes it just gets in the way, but overall, pleasantly surprised. You know, when I looked at the specs, comparing the 15 Pro to the, you know, the iPhone 17 Air, very similar.
Just the camera is a little bit better on the pros, obviously, because you have the third lens and so on. Right. Or the three lenses, so. But again, I just want to keep my life simple and having something light, easy, breezy. It, you know, it has the latest software.
It's been working well for me.
And yeah, I mean, I've. Even a lot of people, I don't think they've opted for the air. So when I go and, you know, show them, like, they're. They're really surprised with the. How much less it weighs compared to their phones.
[00:04:50] Speaker A: Now the question I have is, did you make it humongous with a huge case on it?
[00:04:54] Speaker B: No. I mean, here, I can show you. I got the slimmest case that I could find because I don't trust myself. But yeah, I got a clear translucent case from Apple.
[00:05:06] Speaker A: Or is that a third party?
[00:05:06] Speaker B: No, this is a third party. The Apple one's actually funny enough, like, because I. I got a family member to pick it up from me from the App Store. And so I asked them, like, hey, just get me a case or what have you, and they gave. So it's. It's kind of wild. They have this case. It's like a bumper.
[00:05:26] Speaker A: It's like the iPhone 4 bumper.
[00:05:27] Speaker B: Right. And with the bumper, though, then you got this other thing that, like the strap.
So the bumper with the strap together, it's like $140. I think it was, like, $145. And I. I'm like, okay. Like, I don't know. I tried it for, like a day. I mean, it. It almost seemed like a man purse, but it was, like, really weird.
I'm like, like, the strap was kind of cool, but is it worth, like, $80? I. I don't know. And then the bumper is really not protecting anything. Like, the. The back and front are glass, and so, like, this. I don't know if I really like it.
I mean, the only other option that I was considering was I've seen some on Etsy. There's some leather sleeves that you can go and basically put your phone in. And then, you know, you don't have to actually put a case on it whatsoever. But I feel like I've done that before. I had one. I think it was for my iPhone 6. I had a Hugo Boss case.
And it was.
Even with that one, I dropped the case itself, and then the phone kind of came out and then had some damage around the.
The bezel side. And so I'm like, yeah, I don't know. I think getting. And this wasn't super expensive. The case. I think it was something like maybe 15 bucks off Amazon, but it was the slimmest one. It's translucent. It's clear.
Yeah. Light.
I mean, I would love to keep it kind of just, you know, without anything whatsoever, but I don't.
I don't trust myself. I'm kind of clumsy. So. I remember this was years ago when I got the iPhone 4.
I. I brought it home from Chinook, and I come home and I'm, like, excited to show my family or whatever.
And it slipped out of my pocket as I got out of the car and it landed, you know, smack onto the. The concrete in my garage. And, I mean, nothing happened to the screen, but it. It did get a little bit scuffed. I was somewhat.
So, yeah, I think it. Yeah, it's unfortunate. I don't know what else Apple. Maybe they should. I was even considering getting Apple Care, but then when you read the fine print, it doesn't really do much. It's basically. If there is accidental, you know, damage, basically, you have to. It's almost like insurance. You have to go and give, like, result. Yeah.
[00:07:59] Speaker A: There's a deductible. It's like a hundred bucks or something.
[00:08:00] Speaker B: Yeah. So I don't know. I mean, is it worth, like a hundred bucks every time you drop the phone, maybe?
[00:08:06] Speaker A: I've never dropped a device ever.
I Mean, in a case. But I've never dropped. And I think the only time I ever dropped something and damaged it, I had a Nexus phone and Android, a Nexus 5.
And I had one of those, like, folio cases where you'd open the front, like the front put it to sleep. So it was like totally covered and I was running out to, to help somebody. I was in grad school and it, and it fell out on that face open and it fel.
And I hadn't bought any warranties, not that there was much of an extended warranty that you could buy with a Google Nexus phone at the time.
So I just ended up ordering a whole new phone because it was basically totaled, which was brutal because I didn't have a lot of money at the time. So it was expensive.
[00:08:55] Speaker B: Yeah.
[00:08:55] Speaker A: But, oh, well, live and learn.
[00:08:57] Speaker B: Exactly.
[00:08:58] Speaker A: I typically haven't really dropped or cracked anything. I mean, I did not purchase Apple care with my iPad Pro.
Part of the reason though is it's so expensive. Like, I think the, the, it was, it was like a $250 or something for the extended warranty on the iPad Pro. That was a significant cost. I mean, it was the same cost as like doubling the storage.
[00:09:20] Speaker B: Yeah.
[00:09:21] Speaker A: And so I just thought that's not really worth it. I'll take my chances. Given that I've never, you know, dropped it and I keep it in a case. Plus it has like a little sleeve that I put it in because it goes in my backpack. So I mean, it's like, it's in pretty good shape. But I, I think also, like, my phone sits on a charger 99 of the time. It doesn't really go anywhere. So it's like the likelihood that it'll get damaged is low. I don't use it that much.
[00:09:46] Speaker B: Yeah, well, I think those bigger devices, like, you know, tablets or laptops, it's, it's a lot harder to drop it. Right. Like, I, I don't, I don't. It would be maybe 5% chance if, if that.
But I don't know. With a phone, who knows? Like, even, let's say I'm, I'm showing you something and handing it over, what have you. Or it just slips out of your pocket, like, but anyways. But at the end of it, like. Yeah, I think my advice to people on the phone side is take a look at what you are using the phone for and you know, decide accordingly. And yeah, for myself, I just thought, you know, let's get something like thin light, you know, the screen size. I, I haven't had one this large in a while. I've been sticking to the, you know, the, the smaller pros as opposed to the, the pro max phones. And you know, the weight, it's, it's significantly less. Like it's, you wouldn't think it's a big deal, but yeah, it's, I mean if you're putting it in your pocket or what have you. And I, I agree with you. I guess the, I don't really, especially when I'm like teaching or working. I don't really use the phone that much. I mean I, I still wish, I mean I, I may still do this at some point, but I, I feel like just abandoning a smartphone altogether and just going to like a watch.
[00:11:11] Speaker A: You can do that if you have a cellular connected watch. And you, you could leave your phone always in a bag or even leave it at home because you have to have something, a kind of a base station phone that, that watch syncs to. But you could do that.
[00:11:26] Speaker B: You could. But I, I feel like it's another.
[00:11:30] Speaker A: Additional monthly ongoing cost is the reason I have it. Not so much of the. What I'm afraid will happen.
[00:11:36] Speaker B: Yeah.
[00:11:36] Speaker A: Well.
[00:11:36] Speaker B: And then I think the other thing that my hesitation is imagine, I don't know, like the, my typical. What I usually use my phone for is maybe emailing on the go and I think it would be harder to do it on the watch.
So.
[00:11:53] Speaker A: Yeah. And I don't do any email.
Well, work on any mobile device. Yeah, I'll read my personal email, but I won't just, just to stop myself from checking it more than twice a day.
[00:12:06] Speaker B: Yeah, well, I, I guess, yeah. In terms of. Sometimes I have like time sensitive things and so.
[00:12:12] Speaker A: Sure.
[00:12:12] Speaker B: Yeah. But that's really the only kind of hesitation. Otherwise. Yeah, I would, I would seriously consider it and I, I think you could get around the other thing. Like, I, I don't know. I mean you're saying that you still need.
I think if you had like, let's say if it was you and your wife, you could have, you know, one person has the, the actual smartphone and then the other person has the watch or what have you.
[00:12:39] Speaker A: Yeah, I mean, I think, yeah. I mean you could do that.
[00:12:42] Speaker B: Yeah.
[00:12:43] Speaker A: I mean to me it's like I don't mind having a phone. It's just, it's. But I mean, I can see. I mean I don't. Because I don't use it very much.
And I, I agree with you in terms of the pro phones, there's not a huge difference in, I mean you get an extra lens for telephoto, but I think that's about it. I don't think you're gonna, I don't think taking regular photos, anybody, you would get any better of an image. Maybe there's some other computational photography going on because you're getting a slightly faster chip. I mean, I don't even have the Apple intelligence or Apple non intelligence as I call it, features in the iPhone 15, unlike the one generation behind. And I kind of like that because those things are a real battery drain or they can be.
I did have a question though. So going back to the camera, like, I have know a couple people who have the air and they have said that the single camera is very good.
[00:13:33] Speaker B: I mean it seems to be fine.
[00:13:35] Speaker A: It's excellent.
[00:13:36] Speaker B: Yeah, like, again, like that, that's where like, I, I don't know. I mean, whenever I look at like people who have like the Samsung Galaxy, right? Like, I mean those pictures, I feel like they're way better than even the Apple ones.
[00:13:52] Speaker A: They're a lot more saturated. I feel like they look less realistic.
[00:13:56] Speaker B: Yeah, but I mean I even.
[00:13:58] Speaker A: But it depends on your preference.
[00:14:00] Speaker B: Yeah, exactly. And then I look at, even from a zooming standpoint, like, I don't know, it feels like they're, they're specs. I mean, this is where always, the Android people, they always laugh at Apple people, right? Because a lot of those features, they've already had them for a while or what have you.
Right? But at the end of the day it's a picture. And I mean, look at like these, these photos. I mean, we're, you know, using devices that are how many megapixels? I mean, we never even had this type of 48, you know, computational type of features in the past. Right. So obviously these are going to be way better.
[00:14:38] Speaker A: I mean, I haven't looked at the specs for the latest smartphone cameras, but usually it's like Samsung, the Pixel phones and the, the iPhone Pro or even the regular iPhone are all in the top three and four and it's really like within a hair's distance of who is better. They're good in some things over others. I always get the impression that Apple's phones are the most well rounded. They certainly do video the best, like stabilization and stuff like that. Historically, the iPhone has excelled at video and I think also it's just a little bit simpler to use though for. I would also recommend for anybody with an iPhone who wants to use like power user features or wants a little bit more control.
There's A, a number of like third party camera apps that you can use that are available that give you like you know the ability to like shoot raw photos from the camera sensor and to kind of take out all the auto image enhancements and do everything manually much like you would with an slr. So you know depending on how advanced you want to be you might even be able to get some pretty remarkable results like if you care enough. Yeah, I don't so I don't do that. Like my photos have always been like terrific. They continually astound me what I can do with a phone.
[00:15:47] Speaker B: I think it also like I've been watching some videos.
It also helps with like the technique of how you take the pictures.
Sure. You know, how you position the phone and you know, lighting and other things too.
[00:16:00] Speaker A: Right.
[00:16:00] Speaker B: And so I mean I, I would maybe investigate some of those aspects but yeah, overall again I, I'm happy so far.
You know it'll probably get me through another a couple of years and then we'll see what comes out.
I was just really honestly I was disappoint, disappointed with what they came up with for the design of the, the Pros. Otherwise I probably would have gotten it. I mean there's you know 200 difference in price and yeah, I mean it would have been basically the just you know, the logical upgrade but I just, I couldn't you know especially just I think they over complicated it and the.
[00:16:42] Speaker A: Base model devices are so close to the Pros that I think most people should even get that.
[00:16:46] Speaker B: Yeah, yeah exactly.
[00:16:48] Speaker A: I mean then you still get your double lens camera. You don't need a triple lens. The non pro Apple silicon chips are more than sufficient at this point. The RAM has increased across the whole board. Like there's not a huge number of differentiators and that's why I thought the air was so interesting because from a design standpoint it's more of a differentiator to me in terms of additional features that may or may not be useful on the pro side.
[00:17:14] Speaker B: Yeah, actually I think recently I came across an article that that is apparently one of the biggest problems that Apple is having right now is that their chips are so efficient and you know, powerful that it, it doesn't really make much difference. And this is right across all their products from you know, their laptops, their tablets and phones and so on.
[00:17:39] Speaker A: Well you can certainly see it on the Apple Watch side. So I mean like I bought the Apple Watch SE3 so I bought the cheaper one.
I don't feel a variety of reasons it's like half the price.
But they, they went to the S10 chip there. I don't think there is an S S11.
I should look this up just to make sure. I'm pretty sure it's the S10. Yeah. The S10 so. But the S10 was the chip from last year.
[00:18:05] Speaker B: Yeah.
[00:18:06] Speaker A: So that means the Apple Watch 11, you know they all use the same chip across the whole line in fact. So they all they did was upgrade the cheapest watch to come line with every other watch and they didn't even bother updating the chip with the other devices because there's so much headroom. It's so efficient already.
[00:18:27] Speaker B: Yeah. Even the Ultra 3, it has an S10 chip too.
[00:18:31] Speaker A: Yeah, they're all identical inside. And because of that and because you're not dealing with all the extra health features I still get like on device Siri for asking at things without being, having to have a connection which is somewhat useful especially for like setting timers and doing little things like that because that's all I've ever really used it for and it does those things very well.
And I also don't really care about the larger screen size from a design standpoint. I actually kind of like the SE has that boxier tech field. I find the new ones are very.
Whenever they updated the design of the Apple Watch it looks very pill shaped. It looks very 70s to me and it doesn't look like a piece of technology. It's almost too kind of flat and pill shaped. I don't know how to describe it. So when I looked at them in the store I was like this is much more discreet. It's like a little screen.
You know, I don't need a keyboard on it that's absolutely never ever going to use that.
[00:19:27] Speaker B: Yeah, totally.
[00:19:29] Speaker A: So I, I can see your point.
Picking things that are just simple.
[00:19:33] Speaker B: Yeah, I mean I would even consider it like the, I mean we'll see how, what they come up with for their next lineup or what have you. Apparently they're going to come out with a folding phone but I, I would.
[00:19:44] Speaker A: I would have no interest in a folding.
[00:19:46] Speaker B: I'm not interested.
[00:19:47] Speaker A: I want less phone.
[00:19:48] Speaker B: Like I, I think I would even be interested in a smaller phone but they haven't come out with anything decent.
[00:19:56] Speaker A: Yeah, like I mean to me the smartphone is great. I use it for Apple pay. I can use it to look things up. I think the fundamental problem phones is that it's, it's basically an addiction. And so there's a lot of evidence to suggest that you know, just trying to do things through self control isn't very effective. You have to kind of like, if you don't want to eat crap food, you got to take it out of your house kind of a thing.
[00:20:20] Speaker B: Yeah.
[00:20:20] Speaker A: And I think it's the same with phones. Like I, I block. I have a.
It's like, it's. What's it called? It's called, it's like a social content blocker. I can tell you exactly what it's called here because it's an extension social focus and it just blocks everything except YouTube, basically. Because YouTube's really helpful for me because there's a few channels that I follow. I basically cannot go to the web and open any of those things. Plus I don't have any of the apps installed. I'll go and log in and turn it off when I have to do something, but I just, it just dramatically reduce the usage.
There's no need to check that stuff.
[00:20:57] Speaker B: Yeah, no, exactly.
[00:20:58] Speaker A: And a smaller device that's more constrained is frustrating to use that stuff with. So that's like a disincentive right there.
[00:21:05] Speaker B: Yeah.
I mean, another thing that I do is just not have the, If I do have the app installed, I turn off the notifications and then I delete it off my home screen. So if I ever need to pull it up, like one thing that I've like, especially let's say something like WhatsApp, I mean, there's a bunch of people that I don't know why they, even though I've told them about all the security issues and other things, they still insist on it. And so I just turn it off.
All the notifications, I've deleted it off, you know, my home screens. But if I ever need to, because then they, they'll say, oh, well, I sent you this thing. I can pull it up. I don't have to go and reinstall it. But I, I literally, because it's not on my home screens, I don't even remember that I have it.
[00:21:53] Speaker A: Yeah, yeah, yeah, that's a good tip. Like just shove it into that app drawer.
[00:21:57] Speaker B: Yeah.
[00:21:58] Speaker A: And then once a month I go through the app drawer and I'm just like ruthlessly delete things because it's easy to redownload them.
[00:22:04] Speaker B: Yeah, that's true.
[00:22:06] Speaker A: Should we move on to Time magazine's person, slash, Persons of the Year?
So Time magazine usually picks a person of the year.
I should point out for people that Person of the Year is not an award.
Sometimes, you know, we have to remember that people like Adolf Hitler have been Person of the Year.
It is to demonstrate like, like the most influential person or persons of the year. So sometimes it's good, sometimes it's not so good. And I'm not going to go into the details of who's been listed. I think we can all agree that, you know, Adolf Hitler was bad, so that's a safe one.
But you know, presidents often make it on there for better, for worse.
There's all sorts of persons of the year, but this year's a little bit different.
I think it's a bit of a cop out.
Time magazine, the reason we're talking about this is that Time magazine chose the architects of AI. So multiple are Time magazine's Persons of the year.
So there's a few covers that they have for Person of the Year. Ones like AI being built by Scaffold.
You know, they have another one with the famous picture of the, you know, New York Cityscape with the big girder and the people eating their lunch on it. But instead of the people eating their lunch, it's Mark Zuckerberg, it's the CEO of amd. It's, there's Elon, there's a Jensen, the CEO of Nvidia, there's Sam Altman on his computer, plus some other folks.
So they basically chosen all of the AI architects as Person of the Year. Now personally I think it would have been more interesting to pick one of them and be bold. Yeah, specifically Nvidia's CEO Jensen Hong. Because while there's a lot of like money moving around and all sorts of shady business with regards to like AI financing and you know, we can talk about that really. Nvidia's GPUs are kind of the, the central core to making a lot of the infrastructure actually work in the data centers and locally on computers and stuff like that. So to me that would have been a, a more interesting choice. But the article is super long. It goes into, you know, all the different companies from the chip builders, computing providers. So the chip builders are like Nvidia, amd, et cetera, tsmc. The computing providers are like Microsoft, Google, Oracle and Amazon because they use the cloud infrastructure and then the model builders, you know, OpenAI, obviously anthropic, meta, I guess to avoid including Google twice. They didn't include Google and they included like Elon Musk's Grok, which is pretty funny. I don't really think of Grok ever when I think of AI, but I guess, I guess it's popular enough.
[00:24:58] Speaker B: I do have some, perhaps I do have Students that have used Grok and they kind of like it because it's a little bit more blatant.
[00:25:09] Speaker A: So I feel like GROK is like the unhinged uncle version of AI. Like it's, it's, it. I, I question its data set. I, I think. And then I see some of the other ones like Google's and Microsoft is a little bit too railed for me. So I still feel like chat, GPT and Anthropic probably strike the better balance because they're small enough where they can provide more, you know, out there answers and brainstorm but you know, not so insane that it becomes like a reflection of what Twitter has become. Yeah, but that'd be, that's just my feel from its use.
What's interesting is that they highlight and you pointed this out like the spending on AI.
So they say that the capital expenditures of Meta, Oracle, Google, Microsoft and Amazon. So I'm assuming this is combined, this is based on annualized quarterly spending is at like $427 billion in 2025, which is kind of obscene.
[00:26:02] Speaker B: And that's just 2025. So imagine like, you know, they've been investing heavily since I would say probably 22 onwards. Right?
[00:26:12] Speaker A: Yeah. And then there's peak annual US spending as a percent of GDP is like 1.3% and they kind of break it down by tech. Broadband cables, Apollo project versus other things. So like broadband cables were like 1.2% of GDP.
The Apollo project was 0.8. The interstate highway system in the United states was like 0.4. The Manhattan Project was 0.4. So I mean, AI like totally eclipses these other kind of moonshot projects.
And again, this article is worth reading. We'll put it in the show notes. It talks about some of like people who found companionship to AI and you know, the ethical considerations and downsides. So it's interesting.
[00:26:56] Speaker B: Yeah, I mean one thing that was. We talked about this prior to recording. It was strange that they didn't include Satya Nadella from Microsoft.
[00:27:07] Speaker A: Yeah, I am surprised.
It's an interesting point because Microsoft had a rough time because, and rightly so, they were taken to court for antitrust in the late 1990s.
As a result, I think it really changed the company culture and they were kind of trying to be the second best in anything that they did, including the cloud.
Amazon Web Services, the biggest. And then Microsoft is slowly catching up. Microsoft Azure is very robust, but I think it's because they're always trying to avoid this antitrust. Again, they probably always felt that shadow that they're being perhaps more closely watched though maybe Google is going to feel that again soon since they continually. Well actually it was dismissed largely but some other companies have experienced that kind of antitrust claims already. But Microsoft is really what allowed OpenAI to be successful with their initial and been continual investments. And they have kind of. Their bromance has kind of ended between Satya, Nadella and Saml and particularly after like the whole of him firing and coming back and then trying to hire him and him staying at OpenAI and now using other cloud developers, they had to renew their agreement. So I think Microsoft gets like a larger percentage of maybe not stake at the table but percentage of like earnings and you know, they're really the backbone of ChatGPT and, and you know, I think ChatGPT has like 800 million users or something like yeah, it commands.
I don't think people realize this but these AI tools are not on equal footing. Like I've heard that like Copilot, which is essentially ChatGPT technology, has upwards of like last time I looked was like 14% market share. I don't know, maybe Google is like the same or even more now, but ChatGPT has like 80%.
[00:28:56] Speaker B: Yeah.
[00:28:57] Speaker A: Now I don't know how many of those are paid, probably a fraction of that. But even if it is a fraction, that's a large chunk of change.
Of course they spend a lot more money than they earn. Obviously they're kind of taking the Amazon approach where they just scale infinitely until they own everything and then can start charging a profit I guess is their vision.
But yeah, it is interesting that Satya was not included. I feel like Microsoft doesn't really get enough credit for the work they've done with AI and they've started making their own models and the way they integrated Copilot into their suite with my, you know, Copilot365 was, was really well, is really well done. I mean if in, in my, in my experience it connects very well with OneDrive and being able to create documents and sits in the local apps.
But yeah, for whatever reason they're just kind of overlooked.
[00:29:50] Speaker B: Yeah, but I mean even you know, going back to that, the OpenAI side of things in terms of their revenue, I feel the way that they're going is almost like that plan and tier where they're going after enterprise clients and you know, just charging huge amounts for AI implementations on the enterprise level.
Another thing actually we weren't gonna, wasn't something that we had planned but I just remembered Disney has actually partnered up with OpenAI as well. And so they're investing $1 billion and they've signed a three year deal where they're going to allow OpenAI to use all of their Disney and Marvel characters so to create videos and other things. And so that's, I don't know, I mean that seems kind of a risky thing on Disney's part, but.
[00:30:50] Speaker A: Yeah, risky on their part, but probably trying to get ahead of the curb to where they make an investment and have some control over what's going to inevitably happen. Yeah, I'm assuming that's where they're going.
Another thing that was interesting is that they said when they were in this Time magazine article they're talking about, they have a map of the United States that shows kind of where the spending is. So like a huge amount of spending in terms of not as nearly as much in California as you would think in terms of data centers, way more in Texas and kind of like the east Coast. And that might be more about power.
California is like going to run out of power or water at some point based on their population. So perhaps that's the reason data centers tend to sit in Texas and other places maybe where they have better access to like nuclear or something like that.
But they also point out the cost of the hardware. Right. So they say a single price, the price of a single top tier Nvidia chip for AI can run as high as $40,000.
And more than a hundred thousand of those were used to train the Llama 4 model. So you can kind of start to extrapolate the costs. Plus because of the AI buildouts, there's like the huge RAM shortage for computers. And this is going to, this is now pushing up the price of all other consumer electronics that use ram because so much of it is being diverted to AI usage that it's actually a liability. So now if you want to get a computer in the future, it's unfortunately almost for sure going to cost a lot more to continue with RAM upgrades and stuff because there's such a shortage.
[00:32:21] Speaker B: I don't know if you saw this, I saw in the news, maybe it was like a week or two ago.
You know how we're in Alberta, we're getting a lot of data center interest and there's some that are going up. But there was this one, it was actually a project that was not approved because some farm owners, they, they were concerned about the water aspect. And then you know, this one company which is already operating, it's a data center that's operating Here in Alberta and they were showing how, you know, they're getting a bit of a bad rap because in that particular data center what they're using is air to cool as opposed to water. So it was, it was kind of interesting.
[00:33:00] Speaker A: I mean, well, cold air in Alberta is very available, so perhaps that makes sense.
[00:33:05] Speaker B: Yeah, yeah, exactly.
So anyways.
But yeah, I think this article, it's a good long read and yeah, I would, I think people should, if, if you're interested, you can go and take a look. I think the other one other diagram that was kind of interesting too in here, I believe it was from.
Stemming from another article, but just how people use chat GPT.
[00:33:30] Speaker A: Oh yeah.
[00:33:31] Speaker B: I mean people might be interested in that, but you know, I think tied.
[00:33:35] Speaker A: 28 for writing and 28 for practical GU guidance with the top two categories.
[00:33:41] Speaker B: Now I think maybe the last thing that we wanted to talk about, it's maybe a nice segue is you know, being able to use local large language models. And so.
[00:33:54] Speaker A: Yeah, yeah, well, I mean it's, and it's trivially, trivially easy to do now. I mean with a little bit. I mean there's a great guide on it which, which is written by, I don't know, the person with their background is. But Shep. Brian wrote a guide called Intro to Ollama, which I'll explain in a second, a full guide to local AI on your computer.
And basically there's, there's several, like graphic applications you can download that basically look like a chatgpt. Roughly same idea you can, you know, have. There's like a little chat box, there's a plus sign to upload documents. These local AIs, local LLMs, often called small language models, which you and I had talked about would be more prevalent a couple of years ago when we did a presentation at our faculty association.
You can upload documents to have it summarized. They don't really. I think some of these local models do image generation, but it's not going to be what you would get from a frontier model.
But if you're doing general editing text or brainstorming ideas or what most people do on the free tier of a lot of these frontier models.
And if you're considering privacy in particular and you want to upload stuff that's perhaps confidential or your own personal correspondence and you don't want it going into the cloud, a local model kind of solves that because it's not talking to the cloud.
That model is frozen in time based on when it was last trained. And it's Basically running in memory.
And so you do want a computer with at least 16 gigs of RAM. And 24 is better. The more you have, the better because then you can run models, the larger and larger models. So you know, they have, they're, they're based, their size is based on how many billion parameters. So I have a MacBook Pro that has 16 gigs of RAM. It's an M1 Pro, which is. No, but you know, we're on M5 now. But this is, you know, still, in my opinion, just like crushing anything I could do. And so about 7 to 8 billion parameters is maybe, maybe up towards to 12 would be what you would want. But then if, if your computer doesn't have enough memory or fast enough, then it can become unstable, become less reliable, it can be slow. So you want to make sure that you match a model to run locally with the requirements of your computer, of which there's many. Now, Ollama is one of several chatgpt like interfaces that you can download. So there's a graphic user interface or you can run these in the terminal command like I do because I'm just that kind of person, person.
And basically you can Download Ollama. It's Ollama.com O L L A M A we'll put it in the show notes and you'll download that. When you open the graphic user interface for Ollama, it will have some options of models that you can download. But you know, again, keep an eye on how many that the number plus B. The billions of parameters are tied to each one. But on Ollama's website they have all the different models you can install. And if you're running a Mac or Windows, it's just easier to install these from the terminal command and it'll just pull the model in from the terminal command and once it's installed it'll show up as one of the downloaded models in the Ollama interface or, Or in the terminal command there's some commands you can run and it'll tell you which models have been, have been installed.
So it's kind of a nice way to do that. I mean, it's not going to use any power.
It's, you know, if you're concerned about energy usage in AI and I understand that this is kind of like, you know, does everything that you would need to do for basic tasks without, you know, having any impact on the environment or kind of worrying about privacy and confidentiality.
[00:37:47] Speaker B: Yeah, no, totally. And I again, I mean we talked about this. Now it's Been a couple of years but really this feels like the, you know, the future of how and what is maybe the most practical use of the this type of technology is having it local, you know, focused on particular subset of tasks that you need done.
Whereas these large language models, it's you know, everything for everyone, for every topic. Right. So and again it's look at just this past year what a upwards Almost close to $500 billion has been invested in this technology in just one year and we still don't have any kind of clear path to seeing, you know, even having a break even for these companies.
[00:38:43] Speaker A: Yeah, they're so incredibly expensive to run and you know, honestly like it. It's fun to run these on your local computer. You don't even need to be connected to the Internet. I, I would recommend just to give some advice. Probably Meta so or Facebook, I still call them Facebook. They have some open source models that you can run locally. So there's this called Llama which is, you know, and then Olama is kind of a take on that and they have models that will run in a variety of different parameter sizes. So I use llama 3.18 billion parameters. There's also you know, llama 2 which is really small.
So I don't know why, why we'd use one over the other.
I'm not an expert in all of these. There's a Mistral 7, there's all sorts of little models you can use and what I would actually do do is go to a model, go to chat GPT and you know, like cut and paste like all these different models and be like, you know, this is my computer specs, like which ones would I want to run for these tasks and do a web search and let it, you know, comb through it and answer it for you or something because there's just so many little models you can run and some are more specialized than others. But as a General Purpose 1 I use llama 3.18 billion parameters because it's just, it does everything I need and it's kind of fun, right? Like I know, don't.
As time has gone on, I find these frontier models are more just for if I needed to analyze a huge document or something and then I had to go to the work of anonymizing ahead of time or something. But I don't know that I would use a frontier model anymore for basic stuff. I don't think it makes much of a difference.
[00:40:17] Speaker B: Yeah, no, totally. In fact I feel like just even with the latest releases of the Models, they're over complicating the output.
[00:40:29] Speaker A: Yeah. In fact, you know, we can talk about that. One of the things I've noticed with OpenAI, other than it being just kind of a sycophantic companion that'll just kind of tell me how right I am no matter what. Like I've kind of done this where I'm just like, I need you to tell me that I'm wrong and I just will not do it. Like, it's just like it's.
It's like a dog that'll never leave your side. I kind of tempted to.
I think what I would miss is the image generation because I use it for my blog. But I've heard that if you want the most neutral, realistic writing style help, Claude is a lot better.
And Claude has a much larger context window for larger data sets and documents for analysis.
By far the largest. Well, at least I think so. I don't know how it compares to Gemini 3, but I actually, for a Frontier model that I pay for, I may actually switch over to Claude. I've been. I keep looking over at what people are doing, especially around code and stuff, and it's amazing.
[00:41:26] Speaker B: Yeah.
[00:41:26] Speaker A: And so I'm. I'm super impressed with that.
[00:41:29] Speaker B: Yeah, no, for sure. I mean they definitely chat.
[00:41:31] Speaker A: GPT is like the Kleenex of AI.
[00:41:33] Speaker B: Yeah. But I mean, Claude, it is well known for its ability on the, the coding aspects.
[00:41:40] Speaker A: Yeah. So I think there's pros and cons to all these. I mean we've. I've always recommended. ChatGPT is like, I don't know what I want to do. I want something that does everything.
It's probably the most well rounded. It still does coding very well. Its reasoning models are terrific. It does very good documented data analysis. So it's not at all crippled. It's just, it's. Some of the models are a little bit more weighted to different tasks than others and it's hard to keep up with because the stuff changes. Somebody will be ahead of the game and then somebody else will release something else. I mean, I find ChatGPT 5.2 to be a much bigger, bigger improvement from 5.1 than going from 4, whatever to 5. Yeah, it's really cut down on the stupid stuff that it gives me.
Um, but maybe, you know, that's just anecdotal.
[00:42:27] Speaker B: Yeah. Yeah, for sure.
[00:42:30] Speaker A: Well, that's probably a good place to end it. So we will be recording again and I guess we'll chat again next time.
[00:42:39] Speaker B: Sounds good. Talk to you soon.
[00:42:41] Speaker A: Take care Sam.
