Episode Transcript
[00:00:08] Erik: Welcome to ed Tech examined, a series about educational technology and what you need to know. I'm Erik Kristiansen.
[00:00:16] Kris: And I'm Kris Hans.
[00:00:21] Erik: All right, we're ready to start another episode of the examining podcast. I'm Erik Kristiansen, and I am here with my co host, Kris. Good afternoon, Kris.
[00:00:32] Kris: Good afternoon.
[00:00:34] Erik: Does my mic sound okay? We're all good.
[00:00:36] Kris: Yeah, we're all, we're all good.
[00:00:38] Erik: I actually haven't adjusted it for months, so I hope so. I took the, um, spit cover thing off and I just have the fuzz on the end of it, so we'll see if this makes a difference. That thing was always in my face, so I just removed it.
[00:00:51] Kris: Yeah.
[00:00:54] Erik: Do we want to kick it off by talking about, is it AI hype or do we want to start with mini computers? Or did we want to start with. Well, we have tip stuff that would be at the end.
[00:01:05] Kris: Yeah, no, totally. I don't know. I mean, I guess we could go with the, we can start off with this AI, which keeps happening and they keep talking about all, there's everything just aih these days.
[00:01:19] Erik: So maybe we could talk about. Do you want to start, how about we start with Paul Throtts? Because that's the most provocative and. Sure. And then we can. Because it kind of plays into the hype. So I mean, I have the premium subscription, so I can just kind of explain the article. But basically he wrote an article, so Paul Thorot. So like a, I don't know. For decades he's been covering Microsoft and Windows. He's also a co host of a great podcast at which he talked about this, I don't remember the episode Windows Weekly, but he wrote an article, kind of a provocative one, called I will not pay for AI. And he kind of goes through the history of Microsoft. He talks about kind of their product history, the company's philosophy. Um, but you know, he talks a little bit about what AI is supposed to be versus kind of what they're selling. So one of the things he, he says in this article, kind of the main point he makes and the reason he won't pay for AI anymore. I think he was paying for kind of the premium versions of chat, GPT plus, and Copilot Pro when they first came out. Because of course he's a, he's like a journalist, right? He writes about it. But he said that these chat bots aren't really worth paying for because while they do photos and they do some cool stuff, the real value of these generative AI products, I guess it should be more specific is when they're embedded into a product. So they're not a product themselves, but they enhance existing products. And so his argument is that it's kind of like the MSG of software as a service. It just enhances it, it makes it taste better, even if it's not very good for you, I guess, is what he's going for.
So he's like, I will pay for products and services that use AI to provide real value to make those products better and that come through on their promise, but he won't pay for them. Uh, just, just as chatbots. What do you, what do you think of that?
[00:03:38] Kris: Yeah, I mean, that, that's an interesting perspective I can see especially, I mean if you look at Microsoft strategy, and I think that is a good point of how he went through the history of how they've rolled out their products.
It does kind of make sense. And plus it's a little bit different in this case because they're using us. And now, I mean they've even made it free. Right. So they've, OpenAI has released their latest model and made it available without having to pay a subscription, although there are limits. Like you'll probably get throttled down so that you can't use the latest as much as the people who actually are paying for it. But yeah, I mean, I can see what he's talking about because literally these, these chat bots are going empowering or this generative AI technology, it's powering other applications and maybe for those applications you would go and pay for. And that's probably the more worthwhile side of things. I mean it's even, I don't know if you saw or heard about this, but there's even people who are concerned that OpenAI may not even be around at the end of this year because they're burning through so much cash and Microsoft might not even bail them out where they've given billions of dollars in the past because.
[00:05:06] Erik: Yeah, sorry, sorry, keep going.
[00:05:09] Kris: Yeah, so, I mean, and especially because they're, it's, it's interesting even just from a business standpoint, because they're, they're working around those antitrust laws by first going and investing 49% into this not for profit. And so then, you know, for profit.
[00:05:27] Erik: Arm of the not for profit. Yeah.
[00:05:29] Kris: Like, and then not taking full ownership so they take 1% less. And then on the other side they went and I forget the name of that company, but that Mustafa Soleiman took his whatever he was working on and hired all of them. And now they're part of Microsoft AI.
So instead of acquiring them he just hired a bunch. They just hired a bunch of people.
[00:05:54] Erik: And acquihire as they call it. Yeah. So just to flesh this out, one of the things that he talks about in this article, which was by the way I should say was published on August 15, so it's fairly recent, he talks about Microsoft no longer has kind of a one and done you pay for office. And part of the reason is that they release new versions of Office, new features, but people don't necessarily use them. So in the past they would use these ten year old versions of office because it did everything that they need. So for the standpoint of recurring revenue, which I understand they have Office 365 and in fact I dont even pay very much for it. I think maybe because I have other subscriptions or because Im an educator.
I think the family plan cost me $75 a year. I think the individual 365 cost me $55 a year. Its not a lot of money so it is worthwhile for all the cloud storage and it is a good service.
But he says that, he said it's not the 1990s anymore. He's like not only do we as customers of technology have more and better choices, which there's lots of choices for writing and doing stuff, we also have a mix and match as we see fit. Those who pay for Microsoft 365 for the value they receive as I do, can continue using its constituent parts as they wish, but they can get the functionality provided by Copilot from somewhere else. They're all alternatives and now too many to count, many of which are as good or bad, better than copilot, and less expensive or even free. So he's like it's a different world. He also says later on here he says I'll pay for products and services that use AI and provide real value, but not pay for AI. And he's like, it may seem like a subtle distinction, but it's not. He said Microsoft 365 grew from a modern version of classic Office bundle into an offering of such value that it became a no brainer because it is a very good deal actually. But Copilot for Microsoft is a money grab, an overlay, an overly expensive set of functional add ons that should simply be included at no additional cost with the service he's like I'm already paying every single month for. Instead, Microsoft charges a two to 300% markup for copilot per month. This is Copilot Pro to consumers and business users respectively, while providing only minor advantages spread to too thin across the services offering services offerings like too little butter across too much toast. And I think that's a really good description. So I tried it. I tried Copilot Pro.
You watched me use it.
[00:08:38] Kris: Yep.
[00:08:38] Erik: Because, and it was, I really wanted it to work in a desktop app connected to the Internet and be right there and work in context with my writing. I wasn't using it to generate images or to be a general chatbot. It was like, analyze this and it would get stuck. It wouldn't understand. Perhaps it's gotten better now. I tried it in Excel. I tried to select the data and it didn't understand. It's like, no, I can't do that.
It would have all these hangups even though you were working within one document, it just couldn't handle it. I think hes right.
I think the features that you get from it probably arent worth 20 or $30 a month.
[00:09:23] Kris: Yeah, exactly. And thats where even if you think about with OpenAI, what theyre doing is theyre providing and maybe who even knows what the actual terms are with the Apple intelligence. But giving that infrastructure and giving access to that to these enterprise companies to go and create new applications, thats probably a better business model for them.
[00:09:48] Erik: Yeah. So ill talk about this later as part of my tip. But in a contrast to this, and this is what were talking about today, you and I want to focus on kind of the hype, like where is this business model working? Where is it not working?
Ive talked about the product notion on our podcast before. Its basically a note taking to do project management app. Its a terrific app. Big downside is that it has to be connected to the web. So even if you're using the desktop version, Internet goes out. It's kind of like working in Google Docs.
You can add on AI functionality and I'm not sure if I'm getting it at a discount or whatever. And I think maybe it's like $100 for the whole year.
And then the idea is that you're working in it. You can ask it questions, you can be like, here's my notes. Can you make a to do list based on this product project description? And it'll do it and it'll nest the to do list and it'll do all this cool stuff, but, okay, so let's say it's $99 a year plus maybe you pay for the premium, premium version. I have a little tip later about how to get around this if you're an educator, but it does pretty much everything that I would want copilot to do in Microsoft Word. But copilot for Microsoft Word's almost dollar 30 a month. So I feel like notion is probably charging what they need to cover the data center costs from wherever theyre getting their model from. I dont know if theyre using chat GPT or if theyre using anthropics quad or I dont know what theyre using but it works well it understood my prompts and to me thats a reasonable price if its an add on. I mean hes saying it should be built into things automatically and they should just charge what it costs rather than try to have this multi tiered add ons and stuff. And I get that but to me that seems like a much better value.
[00:11:51] Kris: I'm surprised they don't do more of that. I mean I guess these are like educator discounts too right? Probably. But having doing like a 20, you know I think these companies, a lot of them, they should start doing instead like maybe giving some incentive to even lock in for that year.
So maybe instead of being like $20 a month now it's like $100 for the year or what have you and then you're locked in.
But who knows?
[00:12:19] Erik: It's kind of like hosting right?
Sometimes people pay monthly for web hosting. My Erik Kristiansen.net is hosted on WordPress. I have a portfolio that's CA, that's hosted by a company called Hostpapa in BC. But they're all, if you pay for annually or multiple years at a time, you always get those services, those web services at a discount. I find the whole monthly billing is a bit of an issue. And in fact I just wrote a blog article about this not long ago talking about I'm going to write a series about not suggesting that technology is all worse now, but there's certain things that were, there was advantages to the way we used to do things even ten or 1520 years ago. And one of the things I don't like is the subscription creep and especially like the different tiers and then managing and then this is how much you pay and then they nickel and dime you if you want to add for every extra user per month. And these things add up. And there was a CNET article, it was older, I can see if I can find it because it was linked to my piece, which I can put in the show notes.
I'll just bring it up here because it's pretty simple.
And it said according to CNET. And this article was from, wasn't super old, but it wasn't super new. Oh no, it was April 20 April of this year, April 9, they said that the average us consumer spends approximately $1,000 annually on subscriptions, including video streaming, music streaming, e commerce. So that would include Amazon prime, digital tools, groceries, you name it.
Computer World had a similar article, but that was from 2021. So maybe this has changed as a result of post Covid. Hard to say. And they said that the average us consumer spent $273 a month, or $3,200 annually on subscriptions. That's a lot of money.
So $30 a month is a steepen, steep price, or like $100 a year to use AI to a tool that's actually free for most people. Can you get away with the free tier is a pretty good deal?
[00:14:35] Kris: Yeah, absolutely. But yeah, I mean, this is where part of it you have to understand too. I think for most people like the reason why these monthly subscriptions, just from an accounting perspective, it's easier for them to go and account for it. The revenue is coming in slowly.
I mean, even if you think about it, lets say for instance, like Apple going and selling their iPhones through a telecom they cant recognize, even though they sold that $2,000 phone, they cant recognize the income right away because especially if its on a financing thing. So they have to gradually accrue that income. And so these are things that I think it just makes it easier for them and the market likes it. But you look at it from a customer's perspective. I mean, it was interesting recently because I told you I'm going to be moving and so on. And so I've been looking at just the cable and the Internet service and so on. And I found it really fascinating.
There's these subscriptions that we have like with Netflix and Crave and so on. And, and apparently this was totally, I did not know this whatsoever. And maybe this customer service rep, maybe it was misquoting this or, but you could, if you wanted to, you could change, let's say there's some new season of Game of Thrones that comes out on Crave. You could turn on crave for that couple of months or even like, you could actually even change it on a daily basis. So you've watched that movie, you call them up. Or you could actually, I probably even just do it online. And then you've watched that movie, you're done. Now you switch to the other one, the other premium, you know, so I don't know how many people do this.
[00:16:31] Erik: But what do they charge you a day's worth? They probate for like 24 hours.
[00:16:35] Kris: They're worth the same.
[00:16:36] Erik: Right.
[00:16:36] Kris: So they like, let's say it's $20 for per month, so it works out being the same, but you could swap it out from one to the other. And so I, so I guess maybe people are doing this. I never thought about doing that.
I just, I turn on the valves for both of them, so they're both running and so I don't have to do these things. But.
[00:16:58] Erik: I think.
So there's been this move in technology, so software services to move to subscriptions. And I think these annual or monthly reoccurring costs are going to hit a wall depending on how the economy goes. Like this just doesn't work. In fact, just talking to members of my family, the pillar for subscription entertainment is Netflix. And I, of course, added multiple households, even though it costs money, because it's much cheaper for them to be on there for $7 a month per household than it is to create their own premium account from scratch and stuff like that.
Most of my family, they just kind of reimburse me for the cost, which is perfectly reasonable. So that's the one we pay for the highest tier of Netflix, just 30 something a month. But other ones, it's maybe we're subscribing to Prime. I subscribed to prime because I wanted to watch the fallout series and we had to order some stuff for our kid. And so then it was worth it because we could get it right away and it facilitates returns if there's a problem. But I'm not renewing it next month. Yeah, and so you can subscribe for as long as you can. But I think, again, these companies are always trying to get you on the hook for reoccurrent billing. They make it really hard to cancel.
They want you to forget about it so you don't recognize it on your visa bill. But I was adding up. I remember I added up a bunch of charges, little bits of software here or there. And I made a big list on paper of everything I was paying for because it was during our early days of when we were edtech exam. But I was testing a lot of stuff and it was well worth it. And I kind of added it all up and I was like, okay, these are the ones I want to keep. And I was like, oh, I can save $1,000 a year by canceling all this stuff. Yeah, its $1,000 a year that goes into your retirement fund or all that stuff. It adds up over time. So im wondering if AI is going to go that route. What do you think about we talked about this because we did a presentation on it. But do you think that there's going to be continued growth because Nvidia has had this super growth because they got lucky, because it turns out their graphics cards for their computers are really good for crunching numbers for AI, but the data center costs are so expensive. I had read, and I should bring this up and I did not put it in our pin board. Um, I'll search it in a second, that Microsoft had spent like $19 billion or something in capital for, you know, AI. I will double check this when next time I stop talking. But so it's like a huge capital cost. So it makes sense why they're trying to, you know, nickel and dime people for all these add ons. But is this going to make sense in the long run? Like in the long run? And I mean, like, even, maybe even on a few years, aren't a lot of these AI things going to just be able to run locally on devices?
[00:20:20] Kris: Yeah, exactly. Well, and that's, that's the thing, right? Like, I mean, here everybody's, they're buying into the hype. Everybody in their dog has gotten into this, this whole AI, large language models. And part of the presentation that we talked about was actually there's, and even Microsoft, they are investing in small language models that you could probably run locally because really you don't need. And this is where, even if you look at OpenAI, how they've released their latest models, they have that mini one, they have that 40.
You don't need access to billions of documents. Depending on whatever your task is, you can go and just populate it with the exact information that you need to create that language model the data that you need just for that specific task.
And this is where I think Apple, it'll be interesting to see how they roll it out.
They were in talks with both Google and I, OpenAI. They went with OpenAI to go and power their, what they're calling Apple intelligence. But they're at the same time such.
[00:21:32] Erik: A brilliant move, by the way. I just have to put that in, like calling AI Apple intelligence and then co opting the acronym for your own marketing. I can't believe nobody else figured this out. I guess they're the only company with an A that their company starts with, but still.
[00:21:45] Kris: Yeah, exactly. So, but yeah, I mean, now, like, and especially because of the processing power and so on, and they will likely, this is what I mean, part of what both of us, we've talked about this is that they will likely be able to go and run locally on their device, whether it's an iPhone or one of their Macs, and you'll be able to go. And even just the way that their systems, their OS operates, it operates like a database. And that's where if you go into finder it, it somehow just searches not only all the files, it searches the text within the files. And so imagine you get the AI embedded where it's running locally based off of all your files that you have stored on there. And with their big privacy aspect, they're probably the only company maybe again, this is where they could be.
They're always last to get into it. They look at how everybody else is doing it and then they get into their executing on it. And right now maybe it's just using OpenAI's GPT to go and actually work on theirs, but they are developing their own at the same time. So who knows, maybe that's a strategic move as well.
[00:23:09] Erik: So I did some, that's a great point. They are typically last to the party and then maybe it's more thoughtfully implemented. Um, they seemed a little bit more clear about the privacy and what was on device. Depending on which phone you have, I'll never see these things with my iPhone 15 non pro, but I'm fine with that because I don't know if I even want this on my phone. As somebody who's trying to get away from using their phone, I have a new M four iPad pro, so that's good enough.
[00:23:37] Kris: Yeah.
[00:23:38] Erik: Um, but it's interesting. So, but I'm going to come back to Microsoft. So they had a stock drop. This was July 31. This is from Forbes.
So they had great earnings that were reported in April, but their capital expenditures for AI. So they had 64.7 billion in revenue.
They had $22 billion in net income, but their capital costs for AI surged, according to Forbes, to $56 billion. So I dont know if thats for the whole year or if thats for, that must be for the whole year. It couldnt just be for a quarter. I thought it was way less for the quarter. But thats a huge investment and an absolutely enormous amount of money that theyve had to reinvest probably into their azure cloud services. Right. So they're going to have to recoup this investment somehow and who knows, on.
[00:24:45] Kris: Top of it does that. That probably doesn't include the investment in OpenAI, which was 11 billion too, right?
[00:24:53] Erik: No. Yeah. So that would be separate and I find it interesting. That's why I'm kind of, perhaps I'm misunderstanding the AI market. So I see the investment in AI, I guess Copilot Pro and chat GPT for making images and summarizing your notes isn't there's a lot of AI heavy things that they're investing in that because they're probably getting companies to use Azure, to use AI and stuff like that. So there's a lot more down the line. Maybe that's how notions AI works. They're pulling it from the server so they're getting that money back. But while these companies are making these huge investments, I'll leave Apple out of it because they are doing a bunch of stuff on device at the same time Microsoft is. And we haven't really talked about this, maybe we can talk about it more next time because I haven't looked into it so heavily. It's really marketing this copilot Plus PC.
The MacBook Air is like the computer to beat because it's Apple silicon. It's all based on arm. And now Microsoft finally has really compelling Windows notebooks.
The new Surface laptop and Surface Pro are powered by Qualcomm arm chips and they're really good and they have an NPU and they do all this stuff on device. So I guess I'm trying to wonder where are they just charging people monthly for Copilot Pro and these chatbots while they can? And then they're hoping that most of that revenue will come from the enterprise because eventually most of this stuff must just work on device. Right?
[00:26:35] Kris: Yeah, I mean, who knows? I mean, and then on top of it, like how I mentioned, now I remember the name too. So that inflection AI was what Mustafa Soliman had. So imagine hiring these geniuses. I mean he was one of the heads of DeepMind.
So Google's AI. So now you've just hired like, I don't know, 300 people from huge salaries. So now your op costs have gone up immensely too, right? So. But yeah, I mean they're still figuring it out. And I think one of our articles, it looks like one of the only winners in this is the consultants like McKinsey.
They're going and billing and like crazy for people trying to go and implement these generative AI in their organizations because everybody has to keep up with the Joneses and remain competitive.
[00:27:33] Erik: And it's also funny to me too how thinking again about local computing, and we're going to talk about microcomputing. So I think this is a good segue, the Raspberry PI, which people don't really think of as an AI machine they have the newest version of the Raspberry PI, which is just for people who don't know. It's a credit card sized microcomputer. You can put active cooling on it in a little case. It has USB ports, does everything. I think the fifth generation came out about a year ago and it's just leaps and bounds. Better performance than anything that came before it. So that's much newer than the one that I have.
You can spend $50 to get this computer board with two gigs of ram, or you can spend 60 for four gigs of ram, or six, and then you can get eight gigs of ram for $80. And this is the computer. But they released Raspberry PI, released an AI kit for their microcomputer. So I can send you the link. This is from Kanakit, which is a canadian kind of computing distributor. This is where I typically would buy my Raspberry PI from. They're actually really quite good. Okay. And so I think it's tops means trillion operations per second, right? So that's how they measure. Like if you these copilot plus PCs that are coming from Microsoft, I think they have to hit a certain number of tops. There has to be a certain amount of processing power that the neural engine can do to be able to do this AI stuff locally on device. There's a standard.
This AI extension kit for the Raspberry PI is $70, which is not a lot of money. And it does 13 tops. It does 13 trillion operations per second on top of this credit SAR and you can get these and I have an older version but you can get them with 8gb of ram.
And so im thinking you can get a computer. So for the cost of plus the case, lets say you spent $200 with a case and a cooler and a little sd card to install the operating system and the computer. And the AI thing you could have a legit desktop operating system with an AI extension.
You could run desktop Linux, you could probably do some AI. I don't know enough about AI to do the projects for computer programming. Maybe you can run some of these chatbots locally, you can download an LLM or a small language model and run it on device. But I look at the cost of this stuff and I think either the cost of this stuff is going to go down, so there'll be more money to made on the server side or there'll be less and less reason to do this stuff in the cloud.
[00:30:33] Kris: Yeah, no for sure. And I mean that's amazing. I mean the other articles that we came across, like, some of the things that they were talking about was this one article was just talking about how maybe a mini PC would be better than getting a laptop.
Apple is. They're anticipating with their latest Mac mini, it's going to be the size of an Apple TV. So, you know, the Apple TV, very small compared to, you know, the. Which.
[00:31:08] Erik: It's.
[00:31:09] Kris: It's totally doable. I mean, if you think about your phones are so small, they are literally powerful as a computer. And so if you shrunk this down, you could literally for, you know, a much cheaper, like, fraction of the cost. It could be a much more powerful computer than what a laptop would have.
[00:31:31] Erik: Oh, yeah.
[00:31:32] Kris: All right. Obviously you'd have to go and plug it into like, a monitor or what have you. But, you know, it does give you. I mean, that's something to think about. I mean, if it gets small, like how they are, they're, they're making them much more miniature.
[00:31:48] Erik: I'm thinking this might be the way to go. Either get a MacBook. I mean, at this point, a MacBook air, for me would be more than sufficient.
As they get more advanced. I don't think I'm going to get another pro laptop again. I bought it for the ports. I like my m one pro. It's great. It'll last me a long time. Has lots of ram, but I just don't see the value, the return, even the ports of. If I have to use an extra dongle just to have an air plugged in, that gets me identical performance. Doesn't even need a fan. I'm not doing video editing except for recording this live. We're doing audio. I think that's sufficient. But I think that the mini PCs is a better value because you're getting active cooling, which allows the computer to run hotter for longer.
And the funny thing that this article that you gave us, this is from PC magazine, and they list a bunch of reasons why this is better. They point to a bunch of the AMG chips, like these Ryzen seven cpu's. So these are all X 86, aren't they?
These are not even Arm based computers, I'm pretty sure.
So X 86 is not known for being very power efficient. So it doesn't really matter in a mini PC. But I think a lot of these.
What if you had these new Qualcomm chips in a Windows computer and a mini PC? There is a development kit that you can get that's like these new AI PCs. They basically what's in a laptop shrunk into a mini PC.
Apple's is an ArM based system.
It's probably way more power. It'll probably last longer too. Yeah, I think that's probably the way to go.
[00:33:39] Kris: Yeah. I mean I've mentioned this before. If I was that Apple, I mean it probably would cut into their margins and stuff. But remember how they had the iPod shuffle? Imagine if you had a PC that's that size or like a Mac that was that size, that was literally had like a USB. Now I guess it would be a USB C and I, or it could be an HDMI, I guess, part. But imagine you put that, plug that into some screen and literally this little, small little stick is what you carry around everywhere and that's your device. And it just somehow syncs up with the cloud, the icloud and stuff. I mean that would be amazing.
Especially with, I think for most applications is whether you're a student or just a general office worker, what do you really need access to as just some word processing files or excel files, PowerPoint, that kind of thing?
[00:34:44] Erik: Well, you make a good point. I haven't done a review of this. I suppose I can do a mini one now. I bought a m four iPad pro, the eleven inch again.
And you can plug it into. So it's not quite the size of an iPod shuffle. Bigger but much smaller than a laptop. Right. It doesn't have a keyboard attached. I can plug it in to my monitor, I can turn on stage manager and I can use my external monitor as a proper external display connected to a mouse and keyboard via Bluetooth and open windows and resize them and do everything. I have not tested it. Trying to plug in another webcam or something like that to make it a true desktop setup. But you can plug in external storage to an iPad. I'm pretty sure you could plug it into a dock that's plugged into other stuff and it would all connect. I don't know about a webcam maybe. Um, but it's amazing. Um, it's probably as powerful as my MacBook Pro and that's where I have as much ram. But I mean it's very simple, right? You can dock it into anything.
[00:35:55] Kris: Yeah. Like imagine. And this is where I think these companies, I don't know, they're, when you have so much invested in their existing product line, but I mean it's going to take somebody to go and re envision. But why couldn't you, why couldn't you just have some, this docking station that plugs in and now it's just terminals that you just have a monitor. It's like a five K monitor. You plug in whatever device that you have, whether it's this iPad or this old mini computer, and you can just put it in your bag and you're ready to go. You keep on the goal, right?
[00:36:29] Erik: Well, it's not unrealistic because that's what we've been doing with laptops for a long time. My work setup on a Windows laptop is very similar. I could take it home, I pull it out, I plug one cable into it, and I have display, audio, Internet, external storage, mouse and keyboard, everything.
[00:36:45] Kris: Yeah, totally.
[00:36:45] Erik: And so there's no need to have a laptop. It's interesting that you talk about. We should probably come back to Apple. There was an article that you found from, I haven't heard of. This is 404 media. I don't know this website for a while, but it said, oh, no, sorry, that's a different one. It's from Techcrunch. They talked about Apple intelligence because I guess people have access to Apple intelligence through the iOS 18 beta and iPadoS 18 betas. And so it sounds like it stumbles on swear words and controversial topics.
So they, you know, they have a kind of, all the disclaimers, like, quality may vary, all the places where the AI system gets tripped up, it replaced shitty with crappy, you know, something like that, which is pretty funny.
[00:37:43] Kris: This is, again, like, I think you.
[00:37:45] Erik: Have to keep in mind they all do this.
[00:37:47] Kris: Well, look at, we remember when we were comparing chat GPT, OpenAI chat GPT against Copilot, Microsoft's copilot. These bigger companies, because they are publicly traded, they have like, you know, a reputation. They put these guardrails on. So that's why, like, the chat GPT, because, I don't know, their structure is totally different. We've never seen any type of organization like this where it's a not for profit with a for profit arm and nobody even knows what's going on in that organization. And I mean, to the point where, remember, there was the CTO, she was asked point blank in an interview, hey, for your I video generation, a generative AI tool, did you use YouTube? And she didn't know the answer to it. And they did use YouTube videos to go and power that, right. And so that, you know, ability to go and ask for forgiveness instead of permission goes a long way where you can go and push the boundaries. Right. And so, and that's why I think still to this day, probably, I mean, although we did come across another article of how anthropics Claude is getting to the same level as GPT four. So those are the kind of dominant LLMs out there. But again, as we've talked about, I think the future, it may not be these LLMs we've already seen just it isn't scaling properly because the costs are so intensive and they can't figure out how to go and even cover their costs for running these with the electrical and the water and like the data set centers. And so now they're kind of having to reconfigure and they're probably thinking okay well maybe we don't need these billions of documents, we can just have like this million and it can just sit on this on device, on this smaller little device because processing power has gone up, right. And so yeah I'm not surprised at all that they've had issues with these controversial topics and swears because Apple, its not in their interest to go and start getting into. Imagine if it started going and doing all sorts of obscenities and politically sensitive stuff. They could tank their stock raise and the same thing would happen with Microsoft.
[00:40:14] Erik: Well, I havent played with anthropic, I mean youd probably have to use the pro version to test it. And I think we do have access in Canada.
[00:40:22] Kris: Yeah now we do.
[00:40:23] Erik: So I cant really speak to the quality of it.
I would like to try it.
I wish we could get sponsored to do a test. I really don't want to pay for more of these, but I've used a lot of them and I've dumped text into them. I've asked them to read documents and I haven't been able to find something that works as well as chat. GPT's 4.0.
I mean that's just my personal experience.
So I don't know that it works. I even took handwritten notes and uploaded them to uh, to chat GPT plus. And it was able to do a very, very compelling transcription into text. Like it was incredible. It wasn't something sensitive, it was just something I was taking when I was thinking uh, when it was something I've read. So I don't know, it's hard for me again. And I also think about these subscription services and I think about you build your own functionality and you get used to the interface and I. I wonder if there's something, we've talked about this before. I don't know if there's a lock in with AI, I don't know if it's the same. My experience has been that the competitors I paid for just for testing don't even come close personally.
[00:41:38] Kris: Well, and even like what you just mentioned, that's an interesting application of this generative AI tool. Imagine you have those handwritten notes, and this is where I think that creativity aspect comes in. But taking a picture of that written text and then putting it into, let's say, the new multi model, the 40, and just putting in even a crappy prompt like convert this to text, that's something that probably people, I bet you OpenAI didn't even envision this, that people would do this, that could actually recognize the text and then create the text in, you know, type written content, right? And now you can actually imagine as a student, or even if you're in the workplace and you see some, somebody's notes or what have you, you get them. Instead of transcribing them and typing them from scratch, you can just get this device to do it. Right.
[00:42:38] Erik: That's.
I think there's a perception when I tell people I use it that, you know, I'm prompting everything I do and I'm not thinking anymore. But really what I do is I do all my own writing, I take all my own notes, but sometimes if I want to write a blog article, I prefer to write it myself. I may run sections through it that I'm struggling with to get help with grammar cleanup, but I don't really use it to replicate my writing. But I often take notes and they're kind of all over the place because I don't want to stop during the writing process to organize. I just want to get all my ideas out, especially if I'm just putting out kind of fragments.
I haven't written the full text, the prose yet. Often I'll take that text and I'll add it into chat. GPT I have a couple of prompts that I've saved and I've asked it to reorder it, and I've described how I want the blog post to start. And so this is make it based on the description of what I'm trying to get across and the story that I'm trying to tell.
Provide some suggestions for the order of content based on my notes that I find incredibly helpful because it takes hours and hours to do that manually. Um, and it provides often, I don't know if it's a better organizational framework, but it's. And I don't always agree with it. And sometimes I'm like, eh, I'll do it myself, but it's, it provides a different way of thinking about it.
[00:43:59] Kris: Yeah, well, and again, it's just a, you know, there's certain tasks, I think this is where, like, you're, that's a good point that you brought up, erik. People are misinterpreting how people are using these tools. But, like, for instance, here, I'll give you an example. There was chatting with my research assistant for how we're, we're taking these, or generative AI writing tools and integrating it into the business communication course that I teach. And she was telling me about how her grandma, like, they're doing a reunion of the family. And so she taught her grandma of, hey, we're doing this thing. We're going to go on this trip and so on. And her plan was to go and research on the Internet to figure out, like, all the places that you're going to go and see and blah, blah, blah. And so she showed them this thing, and now, like, the grandma is hooked. Like, she's going and figuring out an itinerary. Maybe some of the things are good, are bad, but, like, how much time did that save?
[00:44:54] Erik: But you can edit it, right? Like, I mean, if you cut and paste 30 suggestions and you keep half of them, all you have to do is skim through the 30 and say, I like this, I don't like that. And people don't think. Also, follow up prompts. They often, I see people use it. They prompt once. Well, I think people are figuring out the nuance of how to describe it. You know, what tone do you want? I mean, you can do this all in one prompt, but they also don't feed it back in and get it to build something even better. So often, if it doesn't give me what I want, I'll be like, I like this, this and this, but I don't like these things. Can you give me better suggestions based on what I do like and it'll come back and it'll do it again and it'll be like, you know, I like that. I have this other text that I decided I want to integrate. Can you do it in the same style? And it remembers kind of the conversation and it goes along. So I think that's the value of those kind of general chat bots. But you have to, and it takes some skill. I'm still learning how to do it.
Uploading large documents and books that I've paid for that are reference guides and creating my own GPTs that only search those texts because I don't want to look through the index and stuff like that. Again, it took me some time, but there's training involved to be able to use it skillfully. I don't think it's so easy. Maybe it'll get easier, but I suspect not. I suspect that. But there'll probably always be some skill required, some precision in writing, good props.
[00:46:22] Kris: Well, and that's where, like, even that one article that I came across, where it was talking about how academic institutions are probably even more important now than ever, because those, those skills, that critical analysis, critical thinking, philosophy, like those liberal arts, like crafting a question, those are things that are really important. And I don't think a lot of people fully like how you're saying. I think they're probably at beginner level for even using these generative AI tools, and they don't even go further than that. And a lot of it is a mindset that you have to take this idea of just constantly. There's going to be experimentation. Some things might work, some might not.
But I think there's really great applications, I mean, especially us, as doing writing.
And for anybody who's a writer, you know, now you don't have to have writer's block. You should never have writer's block anymore, because if you're stuck somewhere, you could just say, hey, I have writer's block. And this is what I'm running into. This is what I'm working on. It'll give you in a minute or less, a bunch of suggestions and so you can keep powering forward. well?
[00:47:39] Erik: I do wonder about writing, though, and I wonder that I do like the suggestions it continues to give me. There is a sort of skill to coming up with ideas kind of out of the blue.
I do wonder if, and you and I talked about this when we presented on AI at our faculty retreat conference. I do worry about kind of the skill on the craft of being able to write things by hand without necessarily thinking about the, you know, relying on some sort of thing to prompt me to think of something.
[00:48:13] Kris: Well, and again, there's so much that we don't even know about just with the human. That's why I always keep, especially my students, I always tell them, we have this AI. But you know what? The most powerful AI is still is the human brain, because even us as humans, we don't even fully understand it. But I mean, think about, and there's proven now, like, there's scientific studies, when you aren't even using your brain, you're going on those walks. All of a sudden, it's somehow you come up with those ideas, right?
[00:48:45] Erik: Oh, absolutely.
Yeah. I think that, taking time away, there's actually a great book I just finished. It's not related to attack, but it talks about some of that in another context that you may find interesting. I guess I can highlight it for people who are listening today. I'll just bring up my day one journal here because that's where I keep my book notes because they're private. Maybe I'll do a blog post about it.
It's by a Zen Buddhist named Haman Sunim, and he wrote a book called when things don't go your way. And he talks about getting out of the house and walking and talking about the value of the calm space between all our ideas. And that's where not only happiness comes from, but new ideas come from. The silence is where they come from, right?
[00:49:45] Kris: Yeah. No, absolutely.
[00:49:47] Erik: I did want to move on to a couple of things in terms of, I guess, related to the microcomputing.
Should we cover anything regarding the article about celebrite about which phones were hackable or which aren't?
[00:50:05] Kris: Oh, was there anything?
[00:50:09] Erik: I'm trying to remember. There was an article we had found that's a little bit of a different.
This was from that 404 media co article by Joseph Cox. So I know that it's separate from AI, but it seemed interesting, but I can't see the whole thing. Subscriber only. I was wondering if there's something in terms of, like, what phones are hacked just as a public service announcement.
[00:50:33] Kris: Yeah, and I can't, I mean, this has been a little bit of time now. Like, at the time, I think we could see more of this article, but it was interesting that they were not able to unlock a bunch of the iPhones. And so, you know, whereas probably the other ones are hackable. And so again, I think that bodes well to the commitment that Apple has to privacy.
And really that's one of the key differentiators. Now, if you think about amongst the big tech companies, they're probably all, if you go back to Google's do no evil or whatever, they're probably all evil to some extent. It's just the lesser of the evils. And I think maybe Apple is probably probably the lesser of all of these big tech. I think if I had to go and rank them, I've done this before in our episodes, it's probably Facebook's probably the worst perpetrator, then it's Google, then probably Microsoft and then Apple.
[00:51:38] Erik: Yeah. Yeah, that's true. That's Microsoft's kind of in the middle somewhere. Well, I guess if that's. That's a good summary. I guess what we could do is we could go on to. I have an anecdote, and then we could talk about that. We'll come back to notion, I guess, is the tip, and then we'll wrap it up. Is there anything else that we want to discuss before we move on to those two things? No.
[00:52:00] Kris: I think we've covered a lot of different topics today, so I'm sure it's a good, nice balance that we've done.
[00:52:10] Erik: So I did want to mention a personal project. I wasn't actually going to mention it today because we talked about the microcomputing. That's why I integrated the raspberry PI stuff. So if people are interested, and maybe I'll write up. I think I'm going to write up a blog post about how I did this. But basically, several years ago from you, I purchased two old computers. I purchased an original Bondi blue iMac from 1998, a voodoo PC from the nineties. And you also had. I bought a Mac Pro, or is it power? No Mac Pro. When it was not intel, it was still PowerPC, huge cheese grater style Mac desktop, aluminum case. And so the aluminum case has been kind of a priority project.
So what I did, and again, I'll write this up and maybe I'll share it in the next show notes. But I have a raspberry PI four, so not a five.
A five would be better, but I have a four with 8gb of ram. So I think it's a quad core processor, eight gigs of ram.
And basically what I've done is that that raspberry PI goes into a plastic enclosure with a fan so it has active cooling. I opened up this Mac MacBook, this Mac Pro case, and I have three m. The Raspberry PI computer with a kind of a velcro three m to the back of it. I have put in a internal spinning drive, a terabyte drive within an enclosure. I've put that in this Mac Pro case, and then I fed the wires through the back. So now I have turned this huge Mac Pro case into a teeny tiny. It has this teeny tiny computer inside, and it runs a web server, or, sorry, a network detached storage from my house that's only available kind of on the local network. And so that's what I decided to do. So I think the whole project.
I don't know what the case is worth. I didn't buy those cases from you for very much, so we'll maybe write that off as part of the cost, but I think the whole cost is maybe $175.
[00:54:27] Kris: Yeah, and I mean that, you know, it's funny because at the time we were actually thinking of using it.
It was part of my company. We had two of them and so we were going to make it into a table.
[00:54:41] Erik: So like a glass top.
[00:54:42] Kris: Like a glass top on top of it and then just use it as a little coffee table. And then I thought about it afterwards and at the time I just wanted to clear out some space in my basement. But I even thought about it and I think it's, you know, kudos to you to go and actually run with it. But yeah, using it either as a network attached storage or I thought maybe you just make it into a big cable box where, where you just, just for all your cable management you just throw it in there. Because that thing. And this is where, I don't know. I mean maybe apple for some reason they just keep going, um, you know, they go big, thick and they go, they start going the opposite and getting as small and thin as possible. But that thing is a beast. I mean it's made out of solid aluminum. It's pretty heavy. That's one of the reasons why even like with my monitors, I sometimes even question like the sure it would be great to get like the new studio displays but they're like two grandd. I picked up these like thunderbolts. You know, I have two thunderbolts, I have one cinema display and I mean the thing's still running great and it cost me like fraction of that.
[00:55:53] Erik: Right.
[00:55:53] Kris: Especially if you get them like pre owned.
[00:55:55] Erik: So are they 1440p? They're not 4k.
[00:55:58] Kris: Yeah they're 1440p but not bad. They're pretty solid. They're heavy, they're heavy duty. They have pretty nice speakers built right into them. I mean there's this new, there's a Samsung five k monitor that you can get. It doesn't even have speakers in it.
So I don't know, there's some trade offs that are happening.
[00:56:20] Erik: So yeah, my monitor is 28 inches, it's a 4k, it's an Acer, it has hdr, it's a decent panel. It's not the highest end, but it's serviceable.
It has a really terrible speaker in Ithoodae but I think I paid $350 for this monitor. So not a lot of money.
[00:56:42] Kris: Yeah.
And I think im just kind of biased because I just like having that apple aesthetic where you have the aluminum. I have my Mac mini right here. Thats aluminum. Its sitting here. I got the. Even my webcam, it actually is a logitech one but it has that same finish and it has a magnet on it so I just attaches right onto the over top of the, the existing webcam that's built right into the Apple display so but you have a good.
[00:57:12] Erik: Setup and you're going to have to move it.
[00:57:14] Kris: I'm going to have to move it soon. So anyways we'll see what happens at that point but yeah, continue with I think it's a great project. I look forward to seeing your blog. You showed me what it looks like right before this so I think it's awesome.
[00:57:30] Erik: Ill take some better photos. Its not a huge, it has a terabyte, theyre slow drives, theyre 5400 revolutions per minute not 7200.
But theyre server NAS network storage drives. I think theyre western digital reds so theyre made for long term longevity. I dont know if theyre still the best but at the time they were very good workhorse drives. They're not going to move data that fast but they'll last a long time and I've had two of them and I've had them for years so I had one lying around that I wasn't using.
I actually had it as a time machine drive for my Mac but I have a solid state external drive that's the same size that's this big and so I just use that. I bought it to take it with me and to take it places and I never do. Everything I do fits on a 64 gig usb stick. I'm like this is silly. So I just formatted that drive so it's a very fast backup, automated backup and then I just use the spinning drive and the network attached storage but it's a terabyte. I don't know what I'm going to store on it.
Could be good for my wife and I to share. I can create new users. The way I did it is that if I do a blog post I can put a link to the instructions or put a link in the show notes for that matter. But I basically installed a Linux operating system called Raspberry PI OS. So actually I can change the monitor input and I can boot right into it. Okay but there's a bunch of terminal commands that are finicky and difficult to get to work.
But basically what you do is you format that drive for Linux and then through the terminal commands you basically say after you formatted it, make this drive available on the network so it can be read and written to.
There's other things you can do with network attached storage if you go to a store and buy a network attached storage, they're much easier. You just plug it into your home router and then it'll just work. You'll be able to see it. You can log in with a password if you like, but they don't really run an operating system. They run some custom software so you can boot into it, like synology or something like that. Yeah, and you can do that with the Raspberry PI. I have in the past, I have had a media center that ran open media vault or one of those kind of customized, very lightweight operating systems. So the way I've done it uses up probably a lot more ram, meaning that it's running a full operating system that can do like Windows level stuff in addition to having this drive sit there. So it's not the most efficient. But the reason I did that is that I have this raspberry PI with 8gb of ram, and there's no way a network attached storage is going to need 8gb of ram. That's a huge amount. So I thought, well, why not try to do both things at the same time? I can use it as a desktop computer and I may try. I was thinking of doing a challenge, or I tried to do a whole day's worth of work using that computer just to see if I could do it.
It has Firefox, it has chromium installed in it. So I can go on the web and I can do everything. I think video conferencing would break it, but then I can use that now and I can run it as a storage device. So I can do both.
Yeah, I did have a tip for today, and I'll make it really quick since I already alluded to it. So I've been using this. I talked about it when we talked about our Zettelkast, an episode. This tool called notion for notes, it's kind of a note taking project management software. I did not know this. Originally I was using the free version, but notion has let me just bring up the pricing.
So most people use notion for free, and you probably don't need to do anything. You can do notes with live collaborative workspaces, seven days of page history, you can invite ten guests, analytics for nothing.
Normally, notion plus is like $10 a month. And then you can get unlimited blocks for teams. You get unlimited file uploads, 30 days of page history. So like version control, you can invite 100 guests, so you can make notes public, and you can invite guests to collaborate on things like Google Docs.
You can sync the databases with third party apps, you can create a custom website and host it publicly on notion. So I can even, I think, share with you. I don't know if it'll work.
The test website I made. So you can't use a custom domain for free, but you can use a custom, oh no, that's not the right one. I'll find it in a second.
Yeah, you can basically host a website through this platform, which is pretty sick, and you can do all this stuff. And now again, normally $10 a month, but if you're a student or an educator, you can actually do it for free. And so there's instructions on notion on how to get the stuff for free as an educator.
And basically you get everything that's included in the plus. It's called plus plan for education for individual students and teachers. I don't know if there's some cutbacks from the regular plus plan, but you can get that for free and then also if you want the AI integration. So I got this plus plan for free and I was like, oh, this is great.
I'm really happy with this. So I can use all these things. So they verified my university address and I can log in with, you have to log in with a university email account and they verify it. That's how they know. Maybe there's another way if they don't recognize it.
I log in and I get a pop up that says, hey, I don't know if it's for the, I think it's just for the first year, so I don't know if I'll do it indefinitely.
If you want to include AI into the notion pricing, they will give you to you for a year, if you buy it for a year for 50% off on the education plan. I don't know if it was specifically on the education plan or if it's a promo, but I'm pretty sure it's specifically if you start. So it'll renew at the full price of dollar, ten per member per month. But if you add it to your workspace, I think you get it for $8 a month. Normally, if you, so what does that cost? Eight times twelve is $96 a year us, by the way. But then if you're on the education plan for the first year, they actually give it to you for a whole year for half of that. So it's $50 us or $47 us. And so I'm going to test this. I haven't tested it at great lengthen. I've heard based on my research and looking at Reddit threads, and stuff that the AI integration. This is what Paul Thurrott was talking about. The chatbots are expensive. You're getting hosed for these prices. There's lots of competitors that do free stuff. The real value is when it's integrated into a project.
You can search your archive. You can work within a document you've already drafted to add content or to do kind of things in context. And so I'm really interested in looking this up. So I'm going to play with it and report back. But those are the deals you can get for notion as an educator or a student.
So that's pretty much it for my. For the tip that I had. I think that's pretty sick. So I'm going to. I'm going to kind of use it as my daily driver for notes, and we'll see.
[01:05:44] Kris: Okay, awesome.
[01:05:47] Erik: Is there anything else that you wanted to add?
[01:05:49] Kris: I don't think so. I think we've covered a lot. It's been good to kind of get back into it. We've been busy for the last little bit, so hopefully we can get things a little bit more regular. But, I mean, we did make this conscious choice. We'll. We'll go and discuss topics that make sense and. And so on.
[01:06:15] Erik: Yeah, well, we kind of release them when we can. Yeah. Yeah. Okay. Well, until next time, thanks for all of you, for listening, and I'll talk to you later, Kris.
[01:06:27] Kris: All right. It's always a pleasure. Thanks.
[01:06:29] Erik: Likewise. Take care.
