Fable 5, Omni, and What Big Tech Announced This Summer (Full Transcript)

[00:00:00] Speaker 1: This week on the podcast, Anthropic releases the new Fable 5 model and then quickly pulls in. We've also had a big summer of tech conferences, including Google I-O, Microsoft Build, and WWDC with Apple. We'll dive into what they announced and what's new. Hey everyone, welcome to the Kevin Strafford podcast, where we discuss all things happening in technology. I'm Kevin, and joining me, I have my co-host, Nick Brasi. It feels like every single day when I wake up and I look at what's happening in tech and AI, there's always so much news going on. It seems like one of the big ones is the Fable 5 announcement of it being pulled back now.

[00:00:45] Speaker 2: Yeah, you want to start by talking about Fable and what is going on with that? You had a chance to try it a little bit before it was pulled, right?

[00:00:55] Speaker 1: Yeah, so just as background, Anthropic has some of the leading models in the space. They had Opus 4.8, and just recently they launched Fable 5, which is their latest groundbreaking model and it is exceptionally powerful. In fact, it's so powerful that the U.S. government says that it's a security concern and they've asked Anthropic to basically block the use of this model until further notice. So that's kind of the latest there. And at least I, personally, I've looked at examples of Fable 5 being used, I've also used it, and it's a big step above what previous models by Anthropic have been able to do. It's more powerful, deeper reasoning, and at least just as kind of a story, so my neighbor works at Microsoft, and he had access to the model before the public release or before the general availability. And he said that it's exceptionally good at finding bugs and vulnerabilities in software. And before Fable 5 came out, Anthropic had the Mythos model, and so they gave access to that model to companies before the general release so they could use that to identify, I guess, problems or vulnerabilities in their software and kind of gave them an opportunity to patch it. And now the Fable 5 model came out, and I think very quickly, from what I hear, Amazon kind of reached out to the U.S. government and different companies reached out requesting that the U.S. government kind of prevents further access to it.

[00:02:38] Speaker 2: Now, I think there's a bit of a nuance there, because I think it was that it was suspended, the government asked Anthropic to suspend it for foreign nationals, allowing it to be used in the U.S., but Anthropic said, well, we can't make it available to some and not others. So they just sort of closed it down for everybody.

[00:02:59] Speaker 1: Right. Which is interesting. Yeah. So there's a chance that, hey, if we're not going to make it available internationally, we're just going to prevent access altogether. And I don't know if it's the type of thing where it's hard to restrict certain markets from accessing or whether it was kind of this principled approach where, hey, if others can't access, we're just going to turn it off for all.

[00:03:19] Speaker 2: We've seen something similar with hardware, and because it's hardware, I think it's very different. But when the EU said you have to put USB-C ports on all devices, and so Apple said, we're not going to make two iPhones, one with USB-C and one with lightning. So they just said, we're going to have to change the whole line. But this isn't hardware. It's not like they're building, packaging, and shipping something. But still, I can see the logic. If the rest of the world can't use it, let's just even everything out across the globe. I think it has some logic to it.

[00:03:53] Speaker 1: Yeah. I've heard that before, especially when you're working at one of these organizations and you have kind of market-specific requirements, it does add a lot of extra overhead to kind of adhere to the regulations in certain markets. But yeah, it seems like the thing where if you can't make it available to foreign nationals, it seems like you'd be able to pretty quickly turn off access to those markets. But then the decision seemed to be, hey, we're going to turn it off for all. From what I hear, I think Anthropic is furiously working with the US government to figure out some type of compromise to make this model available again. I think one thing that's hard is everyone has now had a taste of what this model is capable of. And now pulling it back, you see the power of it, and now you no longer have access to it. It's very hard from a consumer perspective. Yeah.

[00:04:51] Speaker 2: You know, it's interesting. One of our colleagues, Garry Chow, who does videos on your channel as well, we were chatting about this, and he made an interesting point that it builds the myth, right? So if Anthropic builds this amazing model that is so powerful that the US government won't let it be used, now it kind of boosts Anthropic's reputation. So now there's this myth that, wow, they're building something that is unbelievable. So that once they get whatever safeguards in place and they do release it, it's going to have this weird reputation stacked on top of it, right?

[00:05:31] Speaker 1: And what's interesting, too, is I think this follows the playbook that other AI companies have used. Just as an example, with OpenAI, I think when they released, I think it was GPT-2, I think initially they said, hey, this is too powerful to make available to the public. And there's too much of kind of a security risk of this. And now, you know, here we are on GPT-5, and we have all these more advanced models. I mean, this is a few years ago where the same story came up and it helped build that myth around the model. And you also question it because, you know, Anthropic, OpenAI, they have IPOs coming up and, you know, they have a vested interest in really kind of showcasing that, you know, how powerful their models are. So yeah, I wonder how much of this is, you know, actual security risk versus, you know, let's almost build the aura or the, you know, the myth kind of around this model.

[00:06:31] Speaker 2: How much of it is hyperbole? And you know, once we had, you know, GPT-2 has been out for years, now we're on GPT-5. So what was the risk? You know, was that actually real or was that just inflating it? And now we've got these companies that are, they're AI companies, but they're really stock market companies, at least at this point in time. And how much of it is just building that reputation for the stock market? But you actually used, did you use Fable a little bit before it was pulled? And can you tell us about how much hyperbole is there?

[00:07:10] Speaker 1: Yeah. So I think probably the, it's a big step up or a big leap from, I guess, Opus 4.8. And I think now that Fable 5 is no longer available, it feels like, you know, having used all these different models, it feels like some of that power has been taken away. I think that's probably one of the best ways to describe it. It's almost like you had this capability, you had this, you know, this ability, and now you've lost some of that. I think that's the best way to describe it.

[00:07:40] Speaker 2: Even in just the short time you've used it, it feels like it's taken something away?

[00:07:44] Speaker 1: I think kind of what I found is, so just on the side, I'm working on just some projects with coding and the ability to, you know, get to what you're trying to build just felt, you know, more efficient. It just felt more effective. And I think like you've seen this with each generation of kind of model development where, you know, ideally each model, you know, each subsequent model comes out and you feel this additional power that comes from it. And it definitely felt like this was kind of continuing in that, you know, the step function up in terms of just becoming better. And so I think, yeah, I think there's a lot of, I guess, truth to it being an improved model. I guess, Nick, have you seen any examples or have you played with it or kind of what have you seen?

[00:08:37] Speaker 2: I did not get a chance to play with it at all. And, you know, something I wonder about is, you know, we talk kind of broadly about the model and the power behind it, but I was curious about specific types of tasks, not necessarily specific jobs, but you were saying you were working on a coding thing.

[00:08:55] Speaker 1: Yeah. Maybe just to share a specific example. So recently I, and this was not me doing this, but I saw an example of someone who tested But they pulled together a full end to end YouTube video using the Fable 5 model where the model was kind of orchestrating all the different steps. So they wanted, you know, they wanted, and I think the specific person's name is, I think it's, I think it's Nick, Nick Herc. But he basically pulled together an entire YouTube video using the Fable 5 model as kind of the orchestrator. And so it went in and it, you know, it wrote the script it, you know, reached out to other tools to, you know, generate the avatar that would, you know, speak to the video. Then it also went through and it, it leveraged other tools to pull together the motion graphics. And then it, it pulled together the entire video and the video, the video itself, I think just in terms of like token usage and whatnot, I think we're looking at like a $100 video, but you watch this video and it's the quality of it is something that I think a human would pull together in terms of, you know, the, the avatar and the, the talk track and the motion graphics. But I think with previous models, the, the polish and the, just the, the tightness of the script just wasn't there. And this with Fable 5, it feels like it's gotten to that point.

[00:10:24] Speaker 2: And that, that, that's a tough one. I mean, when you talk about quality, I mean, these are impressive technical feats that these models can do, but it's like Jurassic Park. You're so concerned about whether you could, didn't stop to think if you should, I mean, you talk about the quality of that, but what about the quality of communication? I'm from one human communicating to other humans. If you pull the human out of the communication loop, what are we doing? Yeah.

[00:10:55] Speaker 1: You know, when I, when I watched that video, I it's a little scary because you, you see the actual you see the actual person presents alongside this kind of AI generated version. And for many people, it's almost indistinguishable. And so you start to wonder there's, yeah, it's, it's, it's no longer one human communicating to another human. Now it's a kind of a computer that's been trained to predict what comes next, communicating to another human.

[00:11:26] Speaker 2: You say it's a computer communicating with people, but I don't think that's what's happening because I don't think that's possible. I don't think a computer can communicate. It can simulate communication and communication is when I have an idea in my head and I want that idea in your head and have to find a way to transmit it to you. But software can't communicate because software is not a person. It doesn't have a thought that it needs to transmit to another person. It's all simulation of communication.

[00:11:53] Speaker 1: Right, right. Yeah. I feel your concern because with, you know, you think of all the malicious uses that you can use this technology for, you know, just in that video example, you know, the, the misinformation that you could spread using an avatar who looks like a real person and, you know, you can have them, you know, you know, describe a very effective argument. So you look at all the misuses, of course, at the same time, you know, there are also a lot of positive uses like, you know, maybe for training or teaching now, you know, it makes it easier to, you know, distribute high quality content that helps more people learn and kind of educate more people. So I think, you know, with everything like that, there are kind of two sides. I think now the government has kind of leaned more towards the, hey, like, let's take a moment. Let's let's pause and let's kind of evaluate this. What I think is interesting, too, is I think shortly before Fable five came out, I think Anthropic basically said, hey, across the industry, we should pause development of models so we could take a moment to kind of evaluate, you know, kind of where things are right now and, you know, how to how to regulate some of the capabilities of these models. But although this this kind of also follows in a track record where I think other companies have before announced that, like, hey, we should like it's interesting because Anthropic wants to pause development right at the moment when they're on top. And so you question like, hey, is that is that more to their benefit or, you know, is it really for the greater good?

[00:13:19] Speaker 2: So you you said that you saw an incredible example of somebody generating video. You've been using it for coding and coding, I think, is I think there's far less of at least the concerns I have of the morality of it. But I think using these AI models for coding have some incredible power there. And I was curious. I mean, I think in a moment we also want to talk about Google's new Omni model. Do you see a comparison between Omni and and Fable? Yes.

[00:13:50] Speaker 1: What I think is interesting, just industry wide, you have all these companies investing in various models. So, yeah, Google recently announced their their new Omni model. And if if you're not familiar with it, you can basically provide different inputs. So whether it's an image, whether it's a text prompt, whether it's, you know, audio, provide all these different inputs and then you can output, you know, different outputs. Currently today, you could feed in, you know, an image and text and then it'll spit out a video. So that's kind of where it is now. And it's, you know, it's a very impressive video model. Like I can take a video of myself and produce an avatar of me and then I could describe a scene for my video and my avatar appears in that video. And I think some of the other, you know, advantages is it has very realistic physics understanding. And so it really recreates the real world in a very, I guess, accurate way. And so this is a new model. But I think like what we're seeing is, you know, with, you know, Fable 5, the Anthropic is kind of, you know, pushing the boundaries on their model here. Google's coming out with, you know, a more capable model. And so we keep seeing, you know, these more capable models coming out of different companies. But yeah, yeah, I've you know, what's interesting for for a YouTube channel, being able to have something like Omni allows you to let's let's say you need some B roll that represents a scene. You know, traditionally, you'd have to go off with, you know, you'd have to bring your camera or film crew, and you'd have to film some scene. And now you simply kind of describe what you need, and it produces it. So as an example, I did a video recently on I think it was on Gemini Spark, but I wanted to showcase how, you know, when you wake up in the morning, you could get this daily briefing that tells you, you know, what's important during the day. And so if I were to film that, I would have to, you know, maybe I'm lying in bed, and I wake up and then I, you know, look at this, you know, digital screen that tells me what's important for the day. But with Omni, I just described like, oh, you know, my avatar is in bed, wakes up, looks at the screen, and it pulled together the scene, and it looked just like me. I mean, it's just phenomenal what it's capable of doing. I was wondering about that shot, if that's what had happened. And I guess that's exactly what happened. Was that through Omni that you generated that? And so that was that was using Omni. And I think for that specific example, I took my like, as an input, I had text, and then I also had the video of my avatar. And so that was kind of the the multimodal input. And then the output was this video that kind of combined those different elements.

[00:16:29] Speaker 2: And so that's, and you actually, you published a video on the YouTube channel about using Omni step by step in the standard Gemini tool. So it's, is it just you go to the Gemini app or the Gemini web chat, and you choose the Omni model and your plugin in text prompts?

[00:16:44] Speaker 1: Yeah, so with Google I O, they announced, you know, Omni and all these other models, they've done a really nice job of bringing everything together into the Gemini app. And you can just go to Gemini.google.com, that'll kind of drop you on the main landing screen. But yeah, it's over on the left hand side, you click on a video icon. And that basically launches the the prompt field where you type in what you're looking for. And it leverages the Omni model behind the scenes.

[00:17:11] Speaker 2: People who know Gemini, in general, who are already using Gemini a bit, it's really not, there's no significant learning curve, it's just choosing the other model and diving in with more sophisticated prompts.

[00:17:23] Speaker 1: Yeah. And in fact, I think what Gemini has done well is you just go to the main prompt field. And I mean, instead of, you know, clicking on the icon to generate a video, you could just say, hey, generate a video of, you know, Kevin waking up in the morning and looking at this, you know, digital screen with my briefing for the day. And when you type in your prompt like that, Gemini recognizes that, hey, you want to create a video. And then behind the scenes, it knows to use the Omni model. So what's interesting is, because I think like with all these different models, like it's confusing sometimes because it's like, well, hey, how do I use that? Or how do I get to that? And I think what Gemini has done a really good job of is you don't really have to think of like, hey, I have this model or that model. Instead, you know, you go to this one experience and if you want a video, you say, hey, make a video or make an image or give me, you know, write, compose an email. And then behind the scenes, it kind of takes care of everything that it needs to do to get you that.

[00:18:16] Speaker 2: Yeah. I was doing a video recently. It might've been Gemini in side of Google Docs, but of course you can just go into the prompt field and say, I need a picture to put in this document, but there is a menu and there is an option to choose generate image and it has the banana icon. So basically what you're doing is you're setting it to use nano banana and you get a better quality image than if you just put in the text prompt, but it does mean go to that menu.

[00:18:45] Speaker 1: I mean, I think, I think what's interesting, like in your, in your Gemini video and in, in docs, I think what you, I, like at the beginning of the video, you showed there are multiple ways to get to the same functionality. And I think a lot of that is, you know, that you're trying to make it as easy to discover this functionality as possible by putting it in multiple places.

[00:19:05] Speaker 2: I think that's true. I think they're also figuring things out. I mean, there is this standard paradigm of a chat field on the right side of the screen that everybody's kind of settled on. And I think they're all also trying to find some way to diversify it. Like what's a different user interface that somebody might use? I think they're still figuring out what the final form will be on these chat assistants or.

[00:19:27] Speaker 1: Yeah. It's almost a type of thing where, yeah, throw, throw a few different approaches on the wall and really see what sticks and like, you know, maybe you put it in the menu, you have an overlaying button, you have a side pane and yeah, see, see what, see where users gravitate towards. And because I think like, so yeah, so much of it is new still. And I like, what is the best design for how you surface something like this? And I think all the companies are still trying to figure it out. And one thing I've noticed too, is when one company introduces a new way of doing it, you know, all of a sudden you see all the other companies kind of follow suit and, and

[00:20:02] Speaker 2: they introduce a similar approach. And they will not stop updating, which, which means we're constantly putting out videos for the new version and the new interface. It's just, it is a never ending stream. They will not stop updating.

[00:20:18] Speaker 1: You know, I think that's a comment I see on the channel a lot where people say, Hey, you know, especially if a product's been around for, you know, let's say 10, 20 years, especially some of the, you know, let's say things like Outlook or software that's been around for a long time or Excel, like, yeah, when, when you move, it feels like things are always kind of shifting around and, you know, positions are changing. So it could be, and even, you know, for video creation, you make a video and then it changes and it's like, well, now we have to make a new video that showcases how to use that.

[00:20:48] Speaker 2: Do you want to talk more about the, the, there were like four or five different tech conferences just over the past month, you were at Google IO, so you've got a pretty good insight on what happened at, at IO, do you want to talk about some of the other conferences and what's dropped there?

[00:21:03] Speaker 1: You know, it, it feels like, especially, especially in the summer, it feels like that's when you have all of the tech conferences. So we just had a Google IO that was towards the end of May, beginning, beginning of June, there was Microsoft build. And I think just recently we had Apple's conference, WWDC. So there were, you know, lots of different, different tech conferences going on. And I think kind of across the board, this is, this is, you know, these companies opportunity to, you know, showcase, you know, what's new, what's coming up, what are the, you know, what are all the things they're working on? I think Google was, was basically like, Hey, like we're innovating a lot in AI, you know, we have this new Omni model coming out. We also have, I guess, Gemini flash, which is, you know, their, their kind of flagship instant model which produces results quickly with high quality. They also have, you know, different tools coming out like Gemini spark, which allows you to, you know, set up AI agents very easily. So I, when I think of Google is, you know, very AI focused, lots of, you know, AI innovation. And then, you know, shortly after that, you have Microsoft going with Microsoft build. So that was at the beginning of June. And I think kind of the, you know, the, one of the biggest things I noticed with build is they, they've now announced their own, I guess what they call MAI models or Microsoft AI models. And I think, you know, you have, you have Anthropic, you have open AI, you have Google and they all have their own kind of flagship models that they're, they're rolling out. They're getting a lot of mindshare and attention in the space. And Microsoft has, I guess, kind of historically relied on open AI to be its main model provider. And I think they also work with Anthropic to show the, you know, various Anthropic models in their applications. But I think Microsoft is deciding that, you know, that's, that's maybe not the right approach moving forward. And they've decided that, Hey, they need to invest in their own models. And they've, I guess, built a collection of, you know, they have their, their flash model, they have their pro model. And so they have a variety of different models that they're, you know, pushing out now. So I think it's a strategy shift on, on Microsoft's part now, where they're developing their own models in house.

[00:23:23] Speaker 2: What do you think of, I guess, they're calling the SLM models. There is a small language models like Aon where they're, and it's funny, they always announce these things and then people say now they're available, but they're not actually available yet. But, but I guess Microsoft is billing out these SLMs to run locally on computers. Did you get any, any insight on this? Yeah.

[00:23:47] Speaker 1: And you, you know, it's interesting because I feel like in the past with these tech conferences, you mentioned that, you know, they, they announce these different models coming up and then, or they announce things at conferences and then there's a lag between when they announce it to when you finally get it. I think one of the best examples is Apple with, you know, a lot of their AI capabilities, they announced it and then a year goes by and a lot of them haven't been delivered yet. I think with Google IO, when they announced functionality, a lot of them were available that day, which I thought was a massive shift from what they've done in the past. I think with Microsoft here, yeah, there is some of that, you know, announce it and there is a little bit of lag. I think what's interesting with some of these SLM models, I think there's, there's more of a focus on, Hey, can we get some of these models onto the device itself where you don't have to, you know, reach out to the cloud to run these different models.

[00:24:39] Speaker 2: What I'm anxiously waiting to see is, you know, there's all this debate about building more data centers and the environmental impact of the data centers. But if we can get models locally on our devices, at what point does that take down the demand on these huge data centers? And I know it's probably an exponential difference between what a data center can do and what the local computer can do. But I wonder what that shift will look like.

[00:25:03] Speaker 1: You know, yeah. And it's interesting because one thing, like I would guess if you were to look at all of the different AI prompts that people are entering in, of course, some of them are probably, you know, very complex and you need really, you need that deep reasoning and deep thought put into answering them. But I would, I would probably say overwhelmingly most people who are using AI are probably using it to, you know, help compose an email or, you know, write, write a message or, you know, maybe, you know, fix, fix some type of, you know, basic bug. And I think with those, with those types of uses, I think a model on your computer can handle those perfectly well. And so you question for, for kind of what the everyday user is doing with AI. I think just, you know, an LLM directly on your PC is probably sufficient for most use cases.

[00:25:52] Speaker 2: Yeah. It's, it's faster. It's more private. It's more secure. And it's a significantly smaller, it, it, what's the word? It's not diversifies, but it, it spreads out that hardware demand to all of the tiny computers that people are carrying rather than huge data centers.

[00:26:12] Speaker 1: And I think what's, what's interesting though, is like you have other companies too, who are, you know, they're, they're launching models that you could run on your PC. Like Google has the Gemma model, Gemma 4, you have Quen 3, the Phi models. And so you have all these different companies pushing out models that can, you know, run just on your device. And I think in fact, some of them could even just run on your phone too. You know, more and more people, people communicate on their phone and you want to leverage AI. And if you have a model that, you know, could help you clean up, you know, a quick message you're sending someone, that's probably good enough.

[00:26:45] Speaker 2: That's a really good point. And there, there, I think there is a whole world of deploying local models on your devices that I haven't cracked into yet. I need to get more into that and see what the capabilities are there.

[00:26:58] Speaker 1: You know, I used a tool recently and I think it's a great example of being able to run AI on your computer. It's called Comfy UI. We released a video on Friday covering this, but it allows you to run all sorts of different types of models just directly on your computer. You could, you know, you could run image models, you could run video models, you can run, you know, just basic, you know, your text generation models. They're actually, I mean, they're probably not quite at the level of what you would get by, you know, hitting a massive data center and running kind of the cutting edge models out there. But they're pretty good, they're pretty good. And I think for the everyday uses, it kind of does the job. And then like, aside from that, you also have, you know, you have all the privacy benefits where it's on your computer, it's kind of within your control. You can work offline. There are no API costs, and so you get a lot of benefits by just running it on your own.

[00:27:53] Speaker 2: And I think one of the things that some of these companies are working towards now, especially Apple, is the running it directly on your phone. And I think they see that as the convenience level of having it right there in your pocket. And I guess Gemini does that to an extent on Android phones.

[00:28:11] Speaker 1: Yeah, and I think, well, I think Apple has this reputation of just, you know, being very secure and private. And so I think that, like, the decision to put it on the phone, I guess, would make sense there. And Nick, I know, yeah, with WWDC, that just took place. I know you've been following that very closely, but what are some of the big things that you saw there? And what are some of the things that stood out to you?

[00:28:37] Speaker 2: You know, there's one big thing that really sticks out to me. And I'm curious if you've followed this at all, because I was watching it live. And so they did their whole song and dance about Siri AI, and it looks really cool, really impressive. And then they said, it's not going to be available in the European Union, which immediately tripped something in my head. I said, you know, if the European Union has regulations that say that this is not safe, maybe that's a reason why I don't want to use it. But it's actually not, that's really not the framing that's going on here. It's not a security or a privacy issue. It's an antitrust issue. It's a competition issue where the European Union has certain laws that says, you know, a company can't have complete control over what software you run on your device. They want it to be open for competition. So if Apple is releasing an AI system on their phones, but they're not allowing Google, they're not allowing OpenAI to release an AI system on their phones, and that's an anti-competitive issue. Have you been following that story at all?

[00:29:49] Speaker 1: You know, it's a very timely thing for me. I've been reading the book. I don't know if you've read this, Nick, but it's called Enchitification. I haven't read it.

[00:29:59] Speaker 2: Of course, I've heard of it.

[00:30:01] Speaker 1: Right. So I'm kind of reading through the book now. It's a really good read. And I don't think it's not just limited to the software industry. But I think, you know, one thing that's been occurring in many industries is, you know, you have a lot of consolidation. You have kind of fewer, like you look at big tech and, you know, you have a handful of companies that are, you know, leading the way in the AI space. I think in general, like my big takeaway is kind of the more competition, the more competition you have, the better it tends to be for the end user. So I think like having interoperability, allowing, you know, other companies to add their AI into the system, like as an end user, that's probably a good thing to have that.

[00:30:46] Speaker 2: So that's the position of the European Commission. And I think that is, you listen to that, and I actually have the transcript. We could even read it here. But I think that is a really important, solid point. But then let's look at the other point. Let's look at what Apple is saying. Apple is giving unprecedented access to the AI tool into your email, into your notes, into your photo collection. Everything that's on your phone can be accessed by this AI model. And Apple has developed that reputation as being more private and more secure. And it leaves me with the question, do I want to give open AI that level of access to my phone? Do I trust Apple more than I would trust open AI or another AI provider?

[00:31:31] Speaker 1: And I think this is where, although I can see from the European Commission's perspective where it's like, hey, make it interoperable, make it so other companies can plug in as well. Like if Apple's solution is the best and the most secure and the most private, well, then, you know, people should choose that. But I guess as it is now, there is no choice where Apple is your kind of one option, especially if you're already in that kind of garden or in the, I guess what they sometimes refer to as the walled garden. But if you make it open to other providers and they're not as secure, then as a consumer, you have that choice of who you want to go with. So I don't know. I think it's easy to come out and make the argument like, oh, it reduces privacy, it reduces security, so we can't do it. It's like, well, did you try and did you look at how you could make it so others could play in that space too? Like, it's very convenient to say, hey, it's going to risk our privacy and security and then kind of close down the argument there and then no one else is allowed to play within your system versus like, hey, let's investigate to see what it would take to make that possible and still do it in a private and secure way.

[00:32:47] Speaker 2: That's interesting. You're advocating for, let's let the individual consumer make that choice of whether they trust it or not. And this is a minefield because I'm tempted to say that an average consumer doesn't have the insight to choose to recognize the real risks. So I think a lot of people would make a choice that would be inherently insecure and would open up a floodgates of privacy problems because they're not completely familiar with all of the risks. But then does that mean I want to just nominate Apple to be the protector of everybody's privacy? Because I don't, they're not the elected official that gets to protect everybody's privacy. They've just developed this reputation.

[00:33:32] Speaker 1: Yeah. And I think like, so like, it's a, I feel like it's a, you know, a lot of decisions, you know, whether it's, you know, the, the app store, right? So if you're, if you're using, you know, an iPhone or an Android phone in the U.S. like on, you know, in the U.S. you have the, you know, you have the, what is it, the app store on iPhones and that, that's the only place you can go download apps. And so Apple is basically declaring who can and who can't be in that app store. And I guess, yeah, when, when you have a company make that decision, you know, what, what if they decide not to include an app that you think should be included? Like now a company is making that decision on your behalf. So yeah, I don't know. I think like at a certain point, like you're buying a device, like you should have some freedom as to, you know, what you could put on there and what you could use on there.

[00:34:24] Speaker 2: Yeah. And there is a part of my brain that says, well, clearly Apple has the privacy and the security locked down, but am I, am I falling prey to their, you know, am I falling into their spell? Have they tricked me to believe, you know, that they are the, the ultimate arbiter of

[00:34:43] Speaker 1: privacy and security and, and, you know, one, one thing I, one thing I always think back to is and, and there've been lots of different lawsuits over this, but you know, you have the app stores. And so on an iPhone, if you want to download an app, you have to go through the app store. And then you're paying a, you pay a 30% cut of any purchase to Apple. Right. And so I think one of the, so Apple's argument is, Hey, by going through our app store, you get this curated experience. You're going to get apps that are, you know, secure that have, you know, basically gotten our stamp of approval on them. And so you're getting more security, but as kind of a side effect of that, you're, you're paying a 30% cut every time there's some type of transaction in the store. Whereas, you know, you look at you know, more typical payment methods and like usually you're paying like maybe a 2% cut. And so they have a massive, massive you know, cut that they're taking because they have control. And so that, that's why as a, as a user or as a consumer, anytime a company has too much control, like that's, that's a little worrying to me that, you know, with, with too much power and control, you know, all of a sudden you can start charging a 30%, you know, slice of any transaction. And so that's what gets me worried with, you know, with AI, if they're choosing, you know, I think this came up too with with iPhones and the default search provider, where if you have an iPhone, the default was Google. And you know, you, I mean, yeah, you could change it, but it wasn't the default. And so, you know, 95, 98% of users would end up using Google as their search provider. And they basically steered you in one direction. And Google is paying, you know, Apple billions and billions of dollars a year to keep it that way. And so I do think when, yeah, when you would have, when you have a company who's making decisions on, Hey, this is the only kind of solution you could use. I feel like that's limiting for end users.

[00:36:31] Speaker 2: Yeah. And, and clearly Apple is lagging behind in AI products, but could this be the lever that they're trying to use? You know, they've got the control over the whole iPhone ecosystem, that whole walled garden. Could they use that as leverage to kind of push themselves back into competition on all these AI products? Yeah.

[00:36:52] Speaker 1: And I think with Apple, they, I mean, they, they have, they have massive distribution, right? They have, you know, you know, millions and millions of users on iPhones. And, you know, if you, if you make this the default AI experience, then, you know, all of a sudden, you know, overnight you're going to get a massive user base and, you know, whatever AI you offer. As a counterpoint though, I think Microsoft, they also have tremendous distribution, you know, via windows and all the different, you know, Microsoft 365 and all those different products. And, you know, they put copilot everywhere and, and yet I think they have like 15 million users of copilot despite having, you know, hundreds of millions of users on windows and all these other products. And, and so I guess distribution doesn't necessarily guarantee success.

[00:37:38] Speaker 2: I'm tempted to make a prediction here. And if we're still having conversations like this in a year, it'll be interesting to see where it lands, but I think it's going to work for Apple. I think if that's the lever they're trying to use, they'll have a battle with the EU and I don't know how that's going to end up, but I think it'll work for them. And for Microsoft, I think it'll work for them in the business segment, but will not work gangbusters for consumers the way Apple's lever will work.

[00:38:05] Speaker 1: Now I think we'll have to see kind of where things land, but I think you're probably right. You know, I think these companies have so much distribution and, you know, so many, such a large user base already that, you know, they could add this in. I think what'll be interesting though, is it feels like the EU is definitely behind with you know, some of the development of its own AI models. And I wonder if this is a way to kind of encourage more development within their markets of you know, some of their own models like Mistral and others. Possibly.

[00:38:37] Speaker 2: I recently came across, I think it's called EU office. I should get that right before I bring it up, but it's basically a entirely developed and supported, all of the servers are run within the European Union as an alternative to Microsoft 365 or Google Workspace. It does not have an AI component that I know of yet, but it's very interesting to me. You know, a system that is entirely in the European Union, because like you say, they are lagging behind on some of the technology, but their legislation, I think is leading a lot of what is controlling, like that whole USB-C port on all devices. There's going to be a new law that says all portable devices have to have a replaceable battery. There are things that the EU is doing that is protecting consumers that will government how they work across the world.

[00:39:29] Speaker 1: Yeah. No, I think so. And I think especially like in the, I think just across the world, a lot of big companies have amassed a lot of power and kind of a lot of control. I mean, just as an example, I recently switched from an iPhone over to an Android phone and you don't actually realize how locked in you are into kind of an ecosystem until you try to move somewhere else. So I think there needs to be that counter against some of that power to make it easier for consumers. I think the EU is kind of leading in that front. I know having worked at big tech, it's hard then because you have different regulations you have to adhere to. And so it makes kind of development with in the EU, you had GDPR. And so then you have to, it introduced a lot of additional costs to adhere to all those different regulations. I think one kind of argument you always hear is, you have too many regulations, it increases costs and that deters investment. So then you question like in the EU is part of the reason maybe they don't have as many big tech companies because there's too much regulation in place. So I think it's a balance where you want regulation, but not too much regulation. So I think you got to find that sweet spot of where consumers are still getting very good solutions, but then companies can also innovate.

[00:40:49] Speaker 2: You know, there was something else that kind of clicked with me as I was looking at these tech conferences, especially WWDC, because historically WWDC has, okay, now for the next 10 minutes, we're talking about macOS. Then we're going to segue over and for the next 15 minutes, we're talking about iOS. They talk about operating systems. That was not the structure this year. It was, here's this AI product. They did a bulk of time on parental controls, but then everything else was this AI product, this AI product. Here's how it works on the iPhone. Here's how it works on the Mac. That combined with, you know, Google IO and Microsoft Build and even Anthropic's independent announcement, it's like suddenly these things are framed not around operating system updates, but around AI model updates.

[00:41:41] Speaker 1: Right, right. Yeah, I think that's where, I mean, I think now, I think investor, you know, I think just in general people are seeing the kind of transformative nature of AI and how much it's changing. And I think, I think all, like one thing I find interesting anytime I, you know, join a call with a Microsoft or any one of these companies, like all the teams within these companies, they're a hundred percent focused on, you know, AI capabilities and AI functions. And I think there's also just a fear from these companies that if they're not kind of investing in that area, they're going to be left behind. Like I think when you, when you look at Microsoft as an example, I think they're like their whole kind of traditional business could be completely appended by all of these, you know, AI tools. So I, I mean, it's not, you know, it's not a surprise that you're just seeing this highlighted in, you know, every company's big event. And I think especially for Apple, because I think there's this perception that they're behind in AI. Like, I think when you think of the big companies in AI right now, you have Anthropic OpenAI and Google, and then kind of Microsoft, I think is a little behind. But then Apple, like you, you don't really hear about Apple much in any of these conversations. And I, I think this is their desire to change that.

[00:43:00] Speaker 2: I think you do hear about Apple, but as the punchline, because they are so lagging behind, because they announced Apple intelligence and it kind of fizzled out and wasn't particularly interesting. And I think that's kind of put them in a negative situation.

[00:43:16] Speaker 1: So I think, I think now they're at the point where they could talk about it, but they also have to start delivering and they have to start showing value from it. And I think my understanding is they're partnering with other companies to actually provide the model behind the scenes. So I think actually Google is kind of the model provider behind a lot of the Apple kind of AI.

[00:43:36] Speaker 2: Right. Yeah. I don't think they're even building their own model. They're building a front end UI, they're integrating it into their operating systems. But I think it's Microsoft's play for a while. It was, you know, powered by GPT, you know, in co-pilot.

[00:43:51] Speaker 1: Right. And I, you know, yeah, it's interesting. You know, with Apple, it kind of follows a pattern with search, right? Like for the longest time, you know, Apple could have developed its own search engine, but instead of doing that, they, you know, paid Google to use or Google paid them to use, you know, Google search engine. And here, I think Apple, it seems like they're making the decision that like, hey, like we're going to build the front end or the end user experience. And then, you know, Google is going to provide kind of the backend or all the kind of intelligence behind the scenes. And it kind of seems they're following this path. I think with Microsoft, I think, so they kind of started down that similar path where they were relying on open AI, but now open AI is partnering with Amazon and all these other companies, and it's not exclusive to Microsoft. And I think the trouble for Microsoft is, you know, their destiny is kind of in open AI's hands, and I don't think they like that position.

[00:44:46] Speaker 2: Yeah. Do you know when they opened up to anthropic models? Just within the past few months, right? Building anthropic models. Yeah.

[00:44:56] Speaker 1: And I think Microsoft's strategy. Yeah. So they, yeah. If you, if you go to, let's say Excel as an example, you could also select Claude as the model kind of behind the scenes. And I think their kind of strategy has been, you know, we offer these different tools and kind of products and we're kind of, you could decide which model you want to use behind the scenes. The model's the commodity where you just choose whichever model and then it kind of runs in whatever application you're in. And yeah, they've changed direction here. There are all these different offerings. All these different companies have their AI and they all have, you know, different plans that they offer. And like, I found myself, because I want to stay up on all these different models, so I end up having multiple subscriptions across all these, you know, different offerings. So I have my Gemini subscription, I have a Claude subscription, I have an open AI Chattu BT subscription and, and co-pilot and, and yeah, so the subscriptions add up.

[00:45:51] Speaker 2: Yeah. You've got to have all those active as teachers, you know, and to stay as active as we can.

[00:45:59] Speaker 1: And I feel like, yeah, that there's been, there's been, I guess, lots of competition over the pricing of these different models. But yeah, in the, in the consumer space, I think it's like the sweet spots, the $20 per month for one of these plans. It feels like that's kind of the, you know, the, the pro or the plus plan is, you know, starts at 20 a month and then, and then kind of goes up from there based on your, your needs.

[00:46:20] Speaker 2: Yeah. I mean, have you been hearing this buzz recently about how, I guess the, the tokens that people use for different AI services, those prices are starting to creep up and there's enterprise deployments that are starting to see the, the monthly bill going up significantly. And I think for the moment, consumer level subscriptions, like you're talking about are fairly stable. And are we looking at a price war now? Is that what's brewing?

[00:46:50] Speaker 1: You know, I, I, one, one thing I've seen happen is Anthropic they've, they've taken access to Claude code, which was originally in their $20 per month plan. And, and they've removed some of the, the access and the functionality and they're putting it in kind of higher tier plans. So I think like what you're seeing is, I think for $20 it's incredible value that you're getting. And I think in many cases, the value that people are getting exceeds the price that they're paying. Cause I think the actual, like building out all these data centers costs a lot of money. And I think at $20 per month, it's not sustainable. But I think the competition is fierce enough right now. And these companies are trying to build a big enough of a user base where, you know, they're willing to take some losses right now on those plans. But I think what we're seeing is, you know, some companies are removing features.

[00:47:41] Speaker 2: I just, I just grabbed my iPad because I remembered, I saw this article literally yesterday and we haven't talked about this, but I think it's exactly what you're saying. This was on TechSpot yesterday, a $200 chat GPT subscription could cost OpenAI $14,000 if you actually used it to its fullest, fullest extent. So they're, you know, they're selling these subscriptions and they're selling these tokens based on the assumption that you'll never use the full maximum amount of what that subscription is actually making available to you. And it's like, it's, it's an 11% usage that they're seeing on people, you know, or companies using these $200 chat GPT subscriptions. And I think that's, that's basically what you're saying.

[00:48:29] Speaker 1: It's interesting. I think I was just reading today, I don't know the specific company, but there's a class action lawsuit where people took full advantage, and I think this might be anthropic, but people took full advantage of their plan where they used all of the, you know, the tokens and everything else that were promised as part of the, you know, $200 per month plan. And I think the cost is far in excess of 200 and so they didn't get full usage of it. And so now they're, you know, basically putting together a lawsuit to, to kind of I guess push back on that. But I think, yeah, because I, my, my guess is the way these plans work is, you know, you have tons of people signing up, some use it probably not too much, and then you probably have your power users who are, you know, pushing it to the max, who are getting a tremendous amount of usage out of it. And those are probably the, you know, it's kind of like, what is it like bandwidth, right? You have a few people on the home internet where, you know, some people are going to, you know, push it to its max and use up every single gigabyte. And then you're going to have, you know, your, your large base who are probably not using as much. But my guess is, you know, the cost of everyday users probably exceeds what they're charging right now. But competition is still keeping that down.

[00:49:37] Speaker 2: You know, that, that tech spot article, I think I ran across it on Reddit and somebody in the Reddit comments said something like if, if everybody who had the membership at your gym showed up, they would not have the facilities for all of those people. It's a gym membership is based on the assumption that people aren't actually going to use it to their full extent. So it makes me wonder, is this, is this sustainable because clearly it's, it's fairly sustainable when we talk about gyms, but it is a very, very different product. So will we see a situation where it just kind of stabilizes and nobody really uses these to their full extent, or do we see a situation where Anthropic, OpenAI, they're spending so much money on data centers, they've got to start driving those prices up. And do we see a shock in what we're actually paying for these tools?

[00:50:28] Speaker 1: Yeah. And I wonder if this is the type of thing, I feel like a lot of industries, like let's take the streaming industry as an example where you have, you know, Netflix and Peacock and all these different services. I think the strategy there was you start with very low cost plans. You have, you know, you have no advertising and you try to attract as many users as possible. And I think kind of what we've seen now in that industry is, you know, as, as it becomes harder and harder to attract new users, I think prices, you know, and as these companies have built large user bases, now they kind of, the pricing goes up, they, you know, you no longer have all plans that are ad free, you know, you start introducing some ad plans. And my guess is these different AI plans will follow a similar path where today it's incredible value at $20 per month. But I think what will happen is, you know, as these companies build larger user bases, I would guess that, you know, the price will go up or features will be, you know, new features that come out will be on the higher price plans. But I think this is probably the golden era of signing up for an AI plan because the value is tremendous. And I think over time, I just don't think it's sustainable. So I think the, I think either the price will go up or the features will be reduced. There'll be some change or you'll have, like, I think even ChachiBT is an example. On their lowest plan now, they have advertising included as part of that. So I think we'll see this change over time. My guess is it may not be as good for the end consumer in the long run.

[00:52:06] Speaker 2: That's the thing I wonder. I mean, there's this rhetoric of how transformative AI is, and it's going to change everything that we do. But if it gets to a price point where people are not able to afford it, does that transformative nature drop off? And I think what's interesting, too, is certain people, well, then, you know, is it the type

[00:52:29] Speaker 1: of thing where, you know, wealthy people are able to afford, you know, the most feature rich and capable, you know, AI plans. And so they could do phenomenal work. But then someone who's, you know, maybe up and coming or struggling, they don't have the means yet to afford these, you know, higher priced AI plans, do they fall behind? So is it? Yeah. Like, do you have almost, like, two different tracks that you go on, where those who can afford it versus those who can't? Is that is it is it like AI model inequality, right?

[00:53:01] Speaker 2: Or will these big companies who have been laying off people and moving work over to AI find that turns out that the AI is more expensive than paying people? And will that shift in another direction?

[00:53:15] Speaker 1: I think I've heard some companies mentioned that where the AI costs are now exceeding what, you know, an entry level employee would cost. And when you weigh those two, which one's giving you kind of more value for your spend?

[00:53:28] Speaker 2: Yeah. And which one has more value to society? Right. Yeah. Yeah. It'll be an interesting couple of years.

[00:53:36] Speaker 1: Yeah. So, yeah, Nick, Nick, we've we've discussed a lot of different topics today. Of course, there's always kind of so much kind of transpiring in the AI space, so much news going on. So always fun to dig into it. I think what I'd love to kind of wrap on is just some of kind of our recent content that we've pulled together that I think is worth looking at to kind of stay up to speed and up to date on the latest in the space. I think we talked about this earlier, but I recently released a video on the Omni model. So if you're interested in kind of going deeper into that and kind of how that works. Nick, I know you have, I think, Gemini in the Google products, and I think also Claude in some of the Microsoft products.

[00:54:15] Speaker 2: Yeah. Yeah. Of course, I'm thrilled to be publishing videos on your channel. And you put out a list of topics recently that I kind of latched onto because I think there is a lot of value to going into these using the AI tools inside of the document application. So let's go straight to, okay, how do you use Gemini inside of Google Docs? How do you use Claude inside of Word? And I'm hoping to continue on a few more videos like that because I think that's a lot of the day-to-day work for individuals is using these to write better documents, make better spreadsheets, make better presentations. So hopefully I'll do a few more on that track.

[00:54:57] Speaker 1: No, that's fantastic. And yeah, we continue learning about all the latest updates and releases in the tech space. And as we learn about them, we pull together videos and share them with all of you.

[00:55:12] Speaker 2: And also, your interview, after Google I-O, you had an interview with a product manager at Google talking about their new products and their direction. So I think that's a very interesting video worth checking out as well.

[00:55:27] Speaker 1: Yeah. One of the things we're going to continue doing is make sure that we can just connect in the industry and get interviews and be able to talk to the different teams and people who are working on this. So that way, everyone watching here can see firsthand what these companies are doing and kind of how they're thinking about this transformation that's happening. But we're going to keep digging into that for all of you. But yeah, I think with that, let us know, I'd love to hear from everyone listening or watching this, what are things that you want to see in these podcasts? What are topics you're interested in? Would you want to have guests come on the show where we talk to people in the industry? This is our first episode, so probably a few rough edges here, and we're going to continue refining and kind of improving. But we'd love to hear from all of you, just what are things that you would like to see? What are things you're interested in? And we'd love to kind of cater this podcast to what your needs are, and we just want to keep improving this and make it more valuable for everyone watching here. But yeah, with that, really appreciate all of you tuning in, and we look forward to joining you next week for the next episode.