Subscribe
Share
Share
Embed
The Insanely Great Podcast is your weekly mix of tech, AI, sports, politics, and whatever other shiny object grabs our attention. One minute we’re breaking down the future of artificial intelligence, the next we’re debating game-day drama, political chaos, or the kind of random topic that somehow becomes the best conversation of the episode. Smart, funny, and just unhinged enough to keep things interesting, it’s the podcast for curious people who like their insights sharp and their conversations a little unpredictable.
00:00:13 - 00:00:28
Welcome to the Insanely Great Podcast, Episode 6 for Thursday, April 23, 2026. The future of AI, smaller, smarter, and more efficient models.
00:00:33 - 00:00:58
And we're back for another episode. With me is my good friend DJ to talk about everything that's going on in tech. But before we get to tech, how was your week there, sir? DJ My week has been really good. I've been getting into an automation platform called N8n that I self-host here on my own network and do a lot of personal workflows through. How was your week?
00:00:58 - 00:01:22
David Nice. It's been good, good. I thought I had a problem with my NAS server crashing over the middle of the night, but turned out to not be the case. Side related bar, have you seen how expensive NVRM and SSDs have gotten? DJ I've seen some of the memes about RAM, but I didn't know it was also impacting SSDs. It makes sense. David Yes.
00:01:22 - 00:01:49
Also, and maybe you can explain this one, the massively large hard drives that are, you know, 40 gigabytes plus, sorry, terabytes, 40 terabytes plus, they're all sold out. From Seagate to Western Digital, all the models that they made of 40 terabytes and larger, even the 44 terabyte models, there's boom, sold out. Can't buy them.
00:01:50 - 00:02:24
DJ Interesting. Last time this happened, it was, I think, flooding in Thailand because a lot of the drives get made in Thailand. I don't know if it's the same way now, but back then it was. I was just thinking, what would be driving up the sale of the purchase of these large 40 terabyte hard drives? I mean, do you think they just want to put like a bunch of AI models on these things, but don't they, don't they need the speed of SSD? David Not necessarily, not when you're training and things like that.
00:02:24 - 00:03:10
Most of that actually happens in VRAM. So you're, it's not like you're pulling a bunch of stuff from hard drives, but if you need to store the model, store the user data, the sessions, all of that's going to not live in VRAM. So you would need a lot of hard drives and you, in theory, wouldn't need to use SSDs for a lot of that stuff. You would have caches and things like that to help offset the speed difference. Dan Right, right. Which kind of brings me to our first story that I wanted to talk about. Techstartups.com reports that 40% of US data centers planned for 2026 are facing setbacks or cancellations driven by power constraints and equipment shortages.
00:03:10 - 00:03:27
A sign that AI growth is now being throttled by physical supply chains, not just software. David Hmm. I thought that that was pretty prevalent when, what was it, Elon Musk made that data center like a year ago.
00:03:28 - 00:03:53
Dan Yeah. And he was, I listened to him kind of rant about that we should be coating the United States with solar panels all over the place. He said the Mojave desert should just be a giant solar panel providing power for these data centers. David In a sense, he's not wrong. I mean, what else are you doing with that space, right? But we could also put, what's it called, windmills in the ocean, which we do as well. Dan Right, right.
00:03:53 - 00:04:20
Yeah. And actually, Elon Musk was talking about this. He said years ago he made this whole point that the best place to put solar panels is several hundred feet above the water in very sunny places because you get the cool from the water. It keeps the solar panels from overheating and they also absorb the sun. And he said nowhere in the United States did they implement this, but China did the following year.
00:04:22 - 00:04:39
David Interesting. David They have all these pictures of them putting these giant structures underwater to support these huge solar arrays that are right above the water. Keeps them cool enough and they, you know, get a lot, get a lot of, get a lot of sun for the solar panels.
00:04:39 - 00:04:48
Who's the, who's the Nvidia guy? David I know exactly who you're talking about. He wears the leather jacket. I don't remember his name though. Dan Yeah, yeah.
00:04:48 - 00:05:10
David The CEO, right? David Hong, I think? Dan Well, he came, yeah. David Yeah, Jensen. Dan He came out with, yeah, Jensen Wang, Wong. He came out with a statement that said that so far the Chinese AI models are just as good as the American ones. David I can't disagree with him, quite honestly.
00:05:11 - 00:06:07
They, they're right up there. Dan Wouldn't you even possibly say they might be even better because they're maybe running off of less efficient hardware, less, not running off the top AI specs and being every bit as competitive? David I would say this. Most of the models that they're coming out with are maybe 300, 400, like some of the larger ones from GLM are like 600 parameters. Where when you're talking about GPT-5, you're talking about Sonnet and Opus. Now you're talking about, and Gemini, you're talking about trillions of parameters. There is like, we don't really know how they run them, but there is, there are rumors that now to save on energy and be able to run things more efficiently, GPT-5 plus, and I think it was starting with 4.5 on Sonnet and Opus, they're mixture of expert models.
00:06:07 - 00:06:36
Now the experts are, are huge, like 30, 40 billion parameters compared to what you would run locally, but you're still, they're not dense models, which would be more efficient for these companies. Dan Right. Makes sense. Speaking of limited, smaller AI models, Google had their AI conference where they were talking about Gemini and they talked about that they partnered with Apple coming for the new Siri.
00:06:37 - 00:07:07
David Wow. They got to break it for Apple. Dan I wonder if this was planned or part of their negotiation or I wonder, you know, it's just like, hey, we're going to pay you $2 billion. We'll be glad to take your money, but we have to, we're going to break the announcement that we're powering the backend for all the AI on your, on your phone. David I would say that they beat them to the punch, but isn't WWC coming up pretty quick? Dan It's June. David Yeah.
00:07:07 - 00:07:11
Like they're not that much further ahead, but that is a pretty big revelation.
00:07:14 - 00:07:38
So is it, did they say, I didn't catch it, is it going to be local plus cloud? Like you're going to have Gemini local on your phone too? Dan I think there's going to be a small model that's on the phone, on the phone, and it's going to be able to reach out to the, like a private cloud server that's going to interact with a larger model there.
00:07:41 - 00:08:26
David Sort of like now. Dan Yes, it'll, well, I think the goal is for it to be able to interact and do things that they had advertised it could do two years ago. I don't know, make an appointment for me or message my girlfriend, tell her I'm going to be 30 minutes late to dinner or whatever, you know, and in a perfect world, you'd have the AI telling you, it looks like you're going to be 40 minutes late for dinner. I should probably send that text message too. David Yeah. Like truly knows more predictive. Dan Because it has access to your GPS location, knows you haven't left, knows your, what it's already done.
00:08:27 - 00:09:24
You know, when we reach that point with things, I don't know, then we're probably headed in a good direction. David So piggybacking on that, Google also released an app called, I think it was Google AI Studio on iOS, and it was kind of what you expect. You could run the Gemma 4 E2B model, but there was this one little feature that you had to scroll all the way to the bottom to find, and it allowed you to actually control your iPhone, but it ran on its own model. It ran on a model that was about 300 megabytes. So it was maybe four or 500 million parameters, very small to do these very specialized tool call functions. So one of the things that I'm kind of thinking is, it's going to be a mixture of models, but that's Google kind of testing the waters on phone controls for different things, right? On iOS. Dan Yeah.
00:09:24 - 00:10:30
I mean, stuff like that's going to happen and it's going to be a really, really good, you know, step with everything. I think we're just a matter of time before we get better efficiency with everything and the large models kind of become smaller, more efficient, and I don't know, we'll just see where this all ends up. Steve Well, that actually raises two thoughts in my mind. A, one of the stories of this week was some researchers, they found ways to run AI models, smaller ones with this different type of reasoning and they were able to reduce the energy needs for the models overall by 100 times because you could run smaller and they'll run faster too because they're smaller and they put in their study that AI is already consuming 10% of US energy resources and that's only going to grow. So like we need to find breakthroughs like this. But the other thing that was pretty big was this company named Bonsai, like the tree. They trained a model at one bit.
00:10:31 - 00:11:28
So if any of you are aware, the models that you run in the cloud are usually what you call FP16. So that's 16 bits of like, the weights are represented by basically 16 characters. And then you have that, you get eight, your model still pretty smart, but it's starting to lose nuance and then four and then two and then one. But when you get to one, you're literally representing the weights as a one or zero. So what Bonsai did was they trained the model in one bit where everything is just a one or a zero and they can get 8 billion parameters on a roughly 1.7 gigabyte model. And it's just as smart as any of the other 8 billion parameter models, but only at one bit. So then they came out this week with something they're calling, if I can, if I can pronounce it right, ternary.
00:11:29 - 00:11:54
It's basically three, right? The play on that word. It represents weights as one, zero and negative one. So they're calling it 1.5 bit and their models got about 10% smarter just with that change. And the models only got about two or 300 megabytes larger. So you're still dealing with an 8 billion parameter model that is under two gigabytes. So like these could easily run on your phone.
00:11:54 - 00:12:12
They're super fast. Like this is the future, I believe of local inference. You get all the parameters you want. You get a lot more parameters, but it doesn't take a small data warehouse to actually run it for you or, you know, eight GPUs.
00:12:14 - 00:12:15
Right. Right.
00:12:15 - 00:13:20
You know, and wouldn't it, you know, like I just think about like, you know, ways that we can actually use this, like, like I, I think, I think the AI should be able to grow based on the type of person it knows that it's dealing with. For example, if it knows that you're highly into tech and coding and X, Y, and Z, it should be able to download more parameters based on that to make the language models that it might use more efficient on device. But if it needs like a, you know, all this other data that wouldn't probably fit your user type, then maybe it can remove that unless it needs it or is called on to use it. Yeah. So there's two ways to do that. You can either retrain the model every single night on new data and that incorporates the, the new data into the parameters, but that's pretty intensive because to train a model, it's very intensive. But if the model is only a few hundred megs, it's really small.
00:13:20 - 00:13:55
Then you could do that overnight on a phone while it's charging. But most things are going to do it like open call and, and Hermes are doing Hermes agent is the the one out by the Hermes community that does a bunch of open source modeling. It runs this cycle where it makes notes throughout the day on what you're doing. It doesn't write down verbatim what you've done. It makes its own notes about you, which is kind of weird, but it makes sense. Like what it did to help you. And then it'll summarize those into like a daily log.
00:13:56 - 00:14:39
And then every night it dreams where it says, what if this is any significance? And then it stores that in its memory, which is long-term memory. So it's kind of got three phases of memory, short-term, and then that, that middle ground and then long-term memory. And they say that that's more like how humans function. So these, these models are calling, like they keep the session context loaded, but then they pick and choose what they load into context from those three different memory levels. And that's supposed to make them more personalized. And in my findings, like working with these agents, they're getting better all the time and they are becoming more in tune with what I did.
00:14:39 - 00:15:03
Even if I start the session over, it knows like, oh, you were just doing this. Let me help you with this. It's pretty cool. Steve McLaughlin Right, right. So what, what AI models or what AI services do you currently pay for that you've decided are worth a buck? Yeah, I would say right now, Olama's Cloud. Models, they have pretty extensive limits.
00:15:03 - 00:15:44
They have pretty forgiving limits on what you can use. So if you're doing agent at work with some of these tools, you're not going to hit a Anthropic completely closed everyone down. You can use your ChatGPT subscription with these things. ZAI, I used to recommend all the time, but they're actually going from $10 a month to 17 on the low end and 40 on the midterm or mid tier, which is a little rich for my blood for a Chinese model, but their model is pretty smart. And then the other one is Minimax, Minimax 2.7. They have a coding plan. That's 10 bucks a month, really, really high usage limits.
00:15:44 - 00:15:51
Cause it is one of the cheapest models to run. That's really, really smart and really agentically aligned.
00:15:52 - 00:16:20
Wow. So those are the ones that I keep leaning on. Hermes does have their own subscription. You can do 10 or 20 bucks a month or more, but it's more of like a token based thing. Like you can use all your tokens for that one month. Essentially they're just turning like a $20 a month subscription on, but you're paying per token. It's not like the old Llama cloud models where it's a daily and a weekly rolling window.
00:16:21 - 00:16:56
Yeah, that makes sense. Um, I pay for Claude, OpenAI and Gemini Studio. I do pay for Gemini. Yeah. Yeah. But I got the initial version of Gemini with my Google drive upgrade to the one terabyte version. And then I was using the Nano Banana so much with photography that I went ahead and, um, paid for it, which was ridiculous because within two months, uh, Adobe Firefly was just as good as Nano Banana.
00:16:57 - 00:17:21
So then I felt like I, Oh, I also pay for Adobe Firefly. Have you played around with the new image model released from OpenAI? I did. I did. It is very, very good, but yeah. It severely thinks the censorship level on it is extremely high. Oh, I hadn't run into that yet.
00:17:22 - 00:17:45
Yeah. Well, I was making a flyer for, um, a friend's, uh, volleyball camps that he runs during the summer and they're all, it's an all girls volleyball club and he had the wording that he wanted to put in the flyer and the flyer came out great. And then it made all the girls in the video look like they were college professional volleyball players.
00:17:45 - 00:18:27
Look great. Look great to me. And I sent it to, um, uh, the president of the volleyball club and he liked it and said, the volleyball club is mostly for middle, middle school and high school girls. So it's nice that we have these college girls and they are aspiring to play college volleyball. But I really need the girls to be younger and looking like they're having fun. So then I started playing around with chatGBT, uh, image and I said, I've replaced these girls with younger ones, 15 to 17 years old and it immediately said, I can't do that and I, you know, said like, how about 17 year olds? And it said, no, can't do that either.
00:18:28 - 00:19:35
And I was like, all right, well, what can I, what can I do to get around this? And I was actually researching this on Reddit that if you provide a reference point, it doesn't know how old the people are in the reference pages. So then it'll extract, create a 3d model of the people in those pictures and then add that in. So I messaged the CEO of this and I said, do you have any volleyball pictures of middle school girls playing volleyball? And he shot me back a text with some of the girls like posing for pictures together with their parents had taken that the coach was able to get and all this stuff. Then my clockwork took that, put that image as a reference point. And all of a sudden now there's three girls that are 15, 16 years old on the cover of this thing. And all I did was tell the, the CEO, you're going to need to get a, a release, uh, forms, have these girls sign that they know that they're a part of your future advertising so that you don't get sued for this.
00:19:37 - 00:19:49
He was, he was more than happy to do it. Loved the way it looked. But, um, again, the only issue was that CHATGBT did not want me having 15, 17 year old girls in this.
00:19:50 - 00:20:39
Yeah. Ironically, because as a joke to make sure that, uh, like as, as a joke, I, I was seriously testing whether they just didn't like the word 15 or 17 or whatever. So I said, put three 15 year old girls wearing full sweat, uh, full Alaskan snow suits to like, not like I'm sexualizing them in any way or shape or fashion, you know, they're fully dressed head to toe, every inch of them is and even that it wouldn't do. I was going to say maybe burkas, but because then you literally wouldn't be able to tell. You, right. If it's a full burka, you just see their eyes. I wonder then if it understands that it would probably still deny it just based on it, contextually, what it understands is the age you're requesting.
00:20:40 - 00:21:24
Yes. Yeah. Which you think it would, um, not be about the age, but about the action that I'm asking the person to do, um, or, um, what, what, uh, what, what they're, what I'm describing for them to do or whatever. Well, it's funny you raise this because there have been a lot of accounts of people taking photos, like you said, and then asking certain uncensored models to picture them naked or picture them, you know, very scantily dressed. We don't need it. We don't need to say, uh, undisclosed models. We can just say it's Elon Musk models that do this. Yes. Yeah.
00:21:24 - 00:21:51
That was unsavory. There's also, I think some Chinese models that have been jailbroken. That also do it. But I mean, um, Elon, uh, wouldn't go as far as to remove, remove the feature. Um, claims that they premium model. Uh, now if you're a premium mess, a subscriber to Twitter X, that now you get the ability to make anybody naked. I didn't even realize it was only on the free tier before.
00:21:52 - 00:22:10
I thought it was only paid before. Well, this was the kind of compromise cause they were this close to getting the X app completely banned from, from the Apple app store. And I don't know how them saying that this is just a premium feature and how that saves their status in the app store.
00:22:11 - 00:22:14
Yeah. But. Yeah.
00:22:14 - 00:22:34
Cause if you think of it in that way, like every company in the, in that industry could just say, well, you, you have to subs you can download the app, but you have to subscribe to see the pictures and videos, right? You know, they don't allow those kinds of apps in the app store. So how does Elon get away with it?
00:22:34 - 00:23:00
Yeah, I don't get it. But Hey, as long as we're on the news of Apple, you know, Apple had some big shakeup news. They had a new CEO that's going to be announced. That's probably going to take over this summer. Apple named a new CEO and reorganized its hardware group into five major areas. Hardware engineering, Silicon, advanced AI technologies, platform architecture, and project management. Yeah.
00:23:00 - 00:23:18
What do you think of all this? I wonder if some of it is not because of how badly the whole Siri thing was over the last two years, you know, they're there, they have fell so far behind. Google has more intelligent assistants.
00:23:19 - 00:24:05
So does Alexa. Everyone but Apple. We didn't really mention the name, but the new CEO that's taking over is John Ternus. He was the chief hardware vice president, senior vice president of hardware before this. So he's he has spent almost his entire life at Apple other than the four or five years where he was getting a mechanical engineering degree at the university of Pennsylvania. People are speculating that because this is a big hardware guy, that this is kind of a return to the Steve Jobs type at the top, whereas Tim Cook was more of a distribution guy, distribution networks, kind of, he was from the books I've read. He's big on consensus.
00:24:05 - 00:24:40
He would talk with the majority of guys in departments and figure out where things needed to go and then give his sign off on it. Whereas John Ternus is, he's had 25 years dealing with the hardware personally, knows what the channels are. He knows what Apple's wheelhouse is, what their bread and butter is. This could be a real, real good shakeup. We could see a lot more products coming our way from Apple. That would be wonderful. What, what's the last new product other than Neo?
00:24:40 - 00:25:15
I think it was the, the VR, right? The VR and the Neo, which ironically it's looking like we're seeing science at both of these were kind of linked to John Ternus. He was also a big headphone guy. He was in charge of the AirPod Max and the AirPods. He was also kind of instrumental in moving to Apple Silicon. I guess the main guy that, well, I, I, let me, let me rephrase that. John was big on the, uh, the iPhone SoC transition, putting it all on a chip.
00:25:17 - 00:25:44
That was kind of his doing. There's a, another guy that they like to show during the keynotes that's big, uh, I forget his name. He's the big Apple Silicon guy that they had to give up nice big pay bump to keep him around. Yeah. Anyway, but it's good. I don't want to see Apple kind of go in the way that they did with the Intel stuff where they kind of get stuck in the sand and are spinning their tires. And can't, can't.
00:25:44 - 00:25:54
Yeah. I mean, they're beholden to another company. That's, that's exactly what has happened in the PC industry. You can't be any faster than Intel or AMD.
00:25:55 - 00:26:34
Right. Right. So, um, you know, uh, it looks like, uh, Tim Cook is going to be a chairman of the board. So he's not completely, uh, signing off and it wouldn't be a good Apple tenure or without a send off from our wonderful president who, uh, had to thank him for being such an ass kisser. He did deliver him a nice glass award at some point, like right at the beginning of the presidency. Didn't they make that thing? And wasn't it made by like a Marine or somebody that had served, uh, served with him or something?
00:26:35 - 00:26:44
I don't recall that, but I recall it was special made like from the same glass as the iPhone screen or something. Yes. Yeah. It was pomp and circumstance around it.
00:26:45 - 00:27:21
Right. Yeah. I just still kind of find it kind of funny that he still calls him Tim Apple. I knew that was coming to this day. Yeah. His posts that he said, uh, he said, uh, lots of people would be paying other brokers, millions of dollars just to get in touch with me to, uh, um, to fix problems and, uh, he said, now when I get the call from, uh, Tim, I say, wow, it's Tim Apple calling, how big is that? I'm very impressed with myself to have the head of Apple calling to kiss my ass.
00:27:21 - 00:27:25
Hold on. He literally tweeted this? Yes, he did.
00:27:26 - 00:27:48
Yes. I did not see that. Yeah. He said during my five years as president, Tim would call me, but never too much. And I would help him where I could. And he said, sometimes Tim was too aggressive in his ask. He said years later, after three or four big helps, I started to say to people, anyone who would listen that this guy is an amazing manager and leader.
00:27:49 - 00:28:11
And he's had a almost incomparable career. Quite simply, Tim Cook is an incredible guy. He's going to lay it on thick. But yeah, he did mention that, um, did mention that, uh, yeah, he had to call and kiss his ass. Well, he wants to keep those free iPhones coming. I'm sure he has never paid for one.
00:28:12 - 00:28:16
Yeah. Ironically, the son, crap, what's his name? But Don Jr.
00:28:17 - 00:28:28
Don Jr. Has, they have an Android version of the Trump phone that they put out that ironically they're calling it the MAGA phone, but it's made in China. Yeah.
00:28:28 - 00:28:36
So it was 47, right? What's that? It was something to do with the number 47, right? Like they put out their own carrier.
00:28:37 - 00:29:13
Yes. Which really all they did was take a, they, they bought space from Verizon or whoever it was or AT&T, and then they rebranded it just like, uh, so many of these companies do and you get the, whatever's leftover data time. And the cheap knockoff phone that is probably way overpriced. Yeah. Yeah. But yeah, so we're, we're, you know, it'll be interesting to see the direction we're going. My hope, my hope is that we get back to the days of building slightly thicker iPhones.
00:29:15 - 00:29:33
I want them slightly thicker. I want, I want larger batteries. I want better cameras. I want better lenses. I want larger sensors. Seems like I'm asking for a lot. Well, especially considering this last year, they came out with the Air, which is even thinner.
00:29:35 - 00:30:00
Like, I feel kind of like that's the direction they're still headed, regardless of who's at the helm. Even with the pro models? I don't know about there. I could definitely, cause I recall there was between some of the pro models. I forget what generation it was. They actually did get like half a mil thicker or something. I think it was maybe when they changed materials or something like that, but yeah.
00:30:00 - 00:30:37
Yeah, you, you would think that they could get rid, easily get rid of the camera bump just by making it just that tiny bit thicker and give you more battery. Yeah. I want more battery and I guess the EU passed a bit of legislation that by 2028, all batteries have to be with screws and replaceable. You mean to get to the battery or just the battery itself? They can make the case hard as nails, but you could take the battery out. If you can rip it off. You have to be able to remove the battery and make it user replaceable.
00:30:38 - 00:30:52
That's good. We'll see. We'll see where that ends up with things that directly is kind of a slap in the face to the Apple design. There's no way they're going to have a little polycarbonate thing that slides up, up and down on the back of that iPhone.
00:30:52 - 00:30:56
No. Does that count for iPads too? I don't know.
00:30:56 - 00:31:13
I don't know. Cause that'd be even, even more iPads have always been the hardest devices to get into. You have to melt all the glue. You have to buy new glue strips to put it back together. It's awful. Yeah. At least the phones now, like you undo two screws at the bottom and the screen lifts right up.
00:31:13 - 00:31:48
Like that's not as hard as it used to be. Yeah. Let's see if I can find it. Yeah. This just talks about that it's next year and that there is a, this, I guess it also applies to the switch console that it has to have a replaceable. Um, um, but I guess it says that any battery that can maintain an 80% capacity level after 1000 cycles isn't covered by the new rulings. That is, that is, Apple's going to claim that.
00:31:48 - 00:32:08
I think they already do. They already have been. Yeah. I don't know. This breaks up a kind of a good, good point where I'm kind of a weird one. Do you have a, your iPhone charging to a hundred percent fully all the time? I stop it at 90%, but I'm pretty much always around the charger now, so I don't want the battery to degrade.
00:32:09 - 00:32:30
Yeah, I do 80% because I'm like you always around it, damn charger. You know, uh, what does your wife do? A hundred percent, 90? I even told her about the features. She's like, yeah, a hundred percent. A hundred percent it is. She'll like, she, like most people, there's no battery etiquette, right?
00:32:30 - 00:32:49
Like most people don't care about this. She'll take it from a hundred and drain it to it's completely dead sometimes and just like, whatever, like her, her, but the crazy part is we buy our phones at the same time. Even though she does that, our battery health is usually right around the same amount.
00:32:49 - 00:33:01
Yeah. You got to wonder then how good the software is that's recording that or measuring that. Oh yeah. That's a good point. Yeah. I don't know why the iPad doesn't have it either. Yeah.
00:33:01 - 00:34:09
Like I just got this, um, the Aura ring, um, to measure body health stuff. And I was reading in a Reddit, uh, that people highly advised that if you want to maintain battery life on this thing to make sure you never charge it past 80%, that because the battery is so tiny on it to begin with, to make sure that you get the most efficiency out of it, to make sure that it doesn't go past 80%, which is kind of weird because there is not a setting on it to stop the charging of it at 80%. So every, every so often when I put it on the charger and it lasts about a week, uh, I have to constantly look at like the app and see where the charging percentages are. Half the time it takes about an hour and a half for, it's got like a little peg that it sits on and then charges there. Speaking of that, it's rumored that we're going to be getting Apple sunglasses. So they're going after Meta. Yeah. Yeah.
00:34:09 - 00:34:18
Well, Meta's partnered with every major, um, um, eyewear company out there between Oakley and Ray-Ban and...
00:34:22 - 00:34:39
One of my pet peeves, I bought the, uh, Oakley ones, tried them on or I was all set to really like them, took them out with me to the golf course and was recording some video and then I got home and imported the video. All of the video is four by three. Yeah.
00:34:39 - 00:35:02
It's all square. It is. I have the Ray-Bans, the ones just before that, same. What in God's green earth? Why, why do I want four by three? Well, the, I guess the thought prevailing thought is you crop it down to whatever you need it at 16 by nine, but then you have to be very thoughtful of like literally how you're looking at things, right? You don't want to look up to try to get something down there cause you'll crop it off.
00:35:03 - 00:35:20
And the quality isn't, I don't know about in the newer Oakleys, but the quality is not that great. Oh, the newer Oakleys, they made a big deal of saying it's 2K. You know, one, 1000 pixels greater than 10 K, but what they don't tell you is it's not 16 by nine.
00:35:21 - 00:35:30
Yeah. It's not, it's not 1080 by 720 or 1080 by 1080 or 2000 by 2000. It's square.
00:35:31 - 00:35:48
Well, think about the platform it's aimed at, right? Instagram. Even it's in, yeah, but even Instagram's kind of gotten away from the square stuff. Haven't they? You know, well, I think, I think the Facebook reels are vertical. Those aren't, those are six by 19, I think.
00:35:50 - 00:36:09
Anyway, whatever. Like I joined this Reddit group where a bunch of, surprisingly, not really surprisingly, a bunch of photographers and videographers were all complaining about the aspect ratio on this, on this shit, it's kind of like. Why would you create, give us at least the option of it recording 16 by nine.
00:36:10 - 00:36:27
Yeah. Yeah. Because I do remember, I don't use the AI or recording part very often, but it's like way up here and way down here. Like you're not even interested in that part of the video, to be honest. Yeah. Yeah. Yeah.
00:36:27 - 00:36:42
It captures a lot, but I was actually talking about the light aspect of it. Like anytime it's even faintly dark, it just loses everything. Yeah. Are you saying because the sensor is so tiny that they put on this thing?
00:36:43 - 00:36:44
Yeah. Yeah. Yeah.
00:36:44 - 00:37:57
I think they probably went with the cheapest sensor that they could find to put it in a, um, a price point of probably, you know, 50, 60 bucks and then mark it up with the $350 sunglasses and wallah, you know, and have the battery life be like, it'll last 20 minutes. Well, that was the thing that I got kind of ticked off. They are not upgrading the battery life for the Ray-Bans. That's still only, I think four or six hours, but the Oakleys are like eight hours and I'm thinking, hold on, couldn't you just replace some of the stuff in the Ray-Bans, like a new processor or a different battery and give them eight hours if you buy a new version, come on. Oh, I think they, I think they put a completely different chip set in there that's more efficient and, you know, whatever, but Hey, as long as we're talking about AI and Facebook and their meta stuff, did you hear that Mark Zuckerberg gave the announcement to lay off about 8,000 Facebook employees and saying that their jobs are going to be in the hands of AI now? But you know, you like, you realize this is hot on the hills of their Muse model that launched last week. So not familiar with that.
00:37:57 - 00:38:43
So their super intelligence lab or whatever it is, they hired a guy for several hundred million dollars a year to come run their new lab because their Lama four models were awful. Like no one really uses them. Most people use Lama three. That's how bad four was still to this day, but he came on to do super intelligence, so he rebranded the whole lab, they made this Muse model. It's not even their smartest model and it's coming up on numbers. It's not great at coding, but it's really good at other things like creative writing, things that Facebook would care about, but not coding. But to that point, they probably are dogfooding their own model for coding.
00:38:43 - 00:39:15
And I'm sure the next few iterations of it will get better at coding. Hmm. Yeah. I mean, I, I remember reading probably three months ago that they said that there was going to be an AI model that was running on everybody's computer at Facebook that was going to be recording keystrokes and looking at everything employees were doing. And they had said that they were using this to train their AI model. Translation, you're training your replacement. Yeah.
00:39:15 - 00:39:46
Against like completely against your will. Like that's unconsensual. Speaking of that though, there is a, an open source application called Pieces. Have you heard of that? So it is something that you run on your own machine of your own accord. And it, it doesn't record keystrokes, but it records things you do. And it's really helpful because if you are a developer or you're an information worker, it will record all the actions you take conversations you have.
00:39:46 - 00:40:37
And then at the end of the week, like if you have a standup or something to go to, you just say, give me standup notes and it knows everything you did because it saw everything. It records it, secures it, store it, stores it securely and only uses local models for all of that part until you ask it a question, you can have it use local models if you want, but you can also have it go to the cloud. So it's, it's seen as the better alternative to what Microsoft tried to do with, what is it, Windows recall or something like that, because it's all by choice, Windows recall, I want to say was going to be opt out. Are you talking about, you're not talking about copilot. No, well, it was to do with copilot. You remember it was going to like record everything you did. And then they literally had this feature where you could rewind back.
00:40:38 - 00:41:00
There was a few Mac applications, literally it was called rewind and then it became limitless and then they, they actually sold it, the company to Facebook about a year ago and they did the same thing. Maybe that's, maybe that's where it came from. Possibly. Yeah. That'd be amazing if it, if it did come from there.
00:41:01 - 00:41:31
Yeah. They were using it to record their employees. Right, right. So have you been playing any new video games? I just want to mention that sadly in America, because we have literally no worker rights whatsoever, they can, you, companies can just load whatever they want on your computer here. But I've worked in companies that are international. They have to be extremely careful in places like Germany, France, and a lot of other EU countries, because they have laws that protect you.
00:41:32 - 00:41:50
I wish we could get some of those around here. I wonder why, I wonder why we don't here. Do you think that, that companies don't view privacy as that important? I think it's a lot of, they view it as it's the company's data and they want it and they have every right to it.
00:41:50 - 00:42:22
Yeah. I mean, I've, I've worked in security and worked in insider threat. Like you pretty much have to suck in a lot of data about what people are doing to get those signals. You don't keystroke monitor, right? You don't take pictures of what is on their computer, but you have to get all the meta metadata around it. I think of it this way, like the US government gets all the metadata about your phone call, how long it was, who you were talking to, who initiated it, that kind of stuff. Your, your, um, tower you used.
00:42:23 - 00:42:54
Right. Most of the insider threat software, if it's going to monitor locally, takes that telemetry data and then tries to piece it all together and see, is there anything anomalistic going on here that could be malicious or bad, even if the user doesn't know about it, right? They could be compromised. So that is not as invasive, but it's more invasive than say what Germany would allow. Right. So they have to be really careful. If we could get more workers' rights, it would have to start with the States.
00:42:54 - 00:43:11
I'm just pretty well given up on anything happening federally, but more States would have to start to adopt it, which all States now are at will States, I believe, so you could be fired for, you know, your eyes being the wrong. So, yeah, all the hazel people have that going for them.
00:43:11 - 00:43:16
Yeah. Yeah. Way to go. Video games though.
00:43:16 - 00:43:36
Video games. There was this game that came out not too long ago. I think it was called Achilles. It's kind of like a similar to Diablo 4 and I love all the Diablo games. They made essentially a vampire survivors knockoff game. It's an auto shooter. So you run around as initially Achilles.
00:43:37 - 00:43:56
And it auto shoots, you mine stone, you build stuff that kills mobs. I just, it's one of those things. It's timed for 30 minutes. So you, you know what you're getting into. The early missions are actually shorter and it's literally something you can play and eat chips at the same time. Cause you're just running around. You don't have to mash any buttons.
00:43:56 - 00:44:01
There you go. I love that kind of game. There you go. There you go.
00:44:02 - 00:44:49
How about you? Uh, I actually have been, I bought a new Xbox X that I bought at Christmas for a reduced rate and it's been sitting in my man cave that I never set up and, uh, I am thinking about setting it up and actually subscribing to one of the passes, the Xbox gaming passes, which one do you recommend? I've always been ultimate mostly because it comes with the PC edition too, but it also comes with all the games. So I think the tier down only has like half the games, like 400, but if you get ultimate, I think it's seven something or 800 games, right? Not all of them are great, right? Like there's Barbie games and stuff like that. Games no one will ever play.
00:44:50 - 00:44:53
Wait, you said Barbie? Oh, sorry. I got to go.
00:44:53 - 00:45:18
All right. I'm there. Sorry. I got to go. You're a fan, but yeah, ton of games. Not all bangers, but since they bought so many platforms over the last five years, like all the games you love are going to be there, all the Gears of Wars, Diablos, like they've gotten almost every publisher except the ones owned by Sony. Yeah. Yeah.
00:45:19 - 00:45:32
I'm kind of sad to say, but this is, you know, I didn't need this kind of Xbox for this. I'm really into like nostalgia games. They actually have a really great app on there for old video games. Yeah.
00:45:33 - 00:46:08
On Game Pass. Okay. Like on my Nintendo Switch 2, I'll play Super Mario Brothers 2. Oh, you, you pay the, you pay for the subscription? I pay for the Nintendo Switch Pass. And you know, it's kind of ridiculous. You get this Switch that's got 2K graphics on the screen, or you can plug it in, dock it, and you get 4K, and then you're playing Super Mario World 2 from like the 1987 or whatever, and, you know, all the graphics are, as you'd expect, like, I think they upscaled them to like 720 or something.
00:46:08 - 00:46:15
I don't know. I would say that that is a pretty good value add. Is it still $80 a year?
00:46:16 - 00:46:31
Yes. Yeah. That's compared to all the other gaming subscriptions out there. That's a pretty amazing value for what you get. Yeah. Yep. Like they added all the Teenage Mutant Ninja Turtles that I used to play as a kid, all those games.
00:46:31 - 00:46:49
The only one, the only one that I haven't been able to find that I had to run an emulator on my Mac to play is the Castlevania series. I was real. A lot of them are on Game Pass. Are they? Yeah. Great. That might've just sold me for that.
00:46:50 - 00:47:04
Another reason to finally hook up my Xbox. And here's the odd thing too. Almost all the games on Game Pass now have cloud streaming. So you really don't even need that powerful of an Xbox to play it. Like you could have got the S.
00:47:04 - 00:47:32
Thanks. Appreciate that. Or being that I'm well outside my 30 day warranty and it's literally been sitting in a box over here. Like that, you know, just shows how bored I was this Christmas when I saw like $100 off cheaper than normal. And I was like, you know, and I think I'll purchase that cause I'm going to play this one day. And it's just sat around. Well, the crazy part is they jacked up their prices right before Christmas.
00:47:32 - 00:47:54
So the discount you saw was based on, I think they raised prices last year twice. Yeah. Well, it was all the, it's all the wonderful tariffs that we had that made everything more expensive. Yeah. And now the round Sony put all the PlayStation prices in there. And, you know, I think the PS pro, unless you're at Costco is $750.
00:47:55 - 00:47:56
Mm-hmm. Yeah.
00:47:56 - 00:48:04
I think it went up to that 50 bucks. What's that? I think it went up that 50 bucks. Cause I remember it came out. It was like 700.
00:48:05 - 00:48:15
Yeah. Yeah. Well, they make a big deal of it. And that like, if you buy the Costco version, it includes two controllers now. Which the controllers are each like about a hundred bucks.
00:48:17 - 00:48:21
Wow. They're more. Yeah. Cause that Xbox controller is like 60.
00:48:22 - 00:48:30
Is it? Yeah. I've had to buy one or two over the, over the past few years. Cause they, man, they wear out.
00:48:30 - 00:48:34
They break. Yeah. No, the sensors wear out.
00:48:35 - 00:48:48
Really? Huh. On the joysticks. You think that would be something that's like covered, but yeah. Who's going to spend the time to send in a joystick when you could just buy another one for 60 bucks? Yeah, exactly. Yeah.
00:48:48 - 00:49:16
And it takes years for that to happen. Yeah. You ever watched, you watch anything on TV, go to the movies at all? I, I think we already talked about this. I tried to watch the fallout TV series, but a trend that I guess other people have noticed I'm coming around to it late is it seems like all the streaming stuff is every episode is like an hour long, no matter what show it is. Yeah. And for you, that's too much of an investment of time.
00:49:18 - 00:49:33
Well, with a three-year-old. Yeah. But I'm kind of used to like the 20, 30 minute episode limit, you know? You want that 70 show with 20, 30 minutes, you're done. Whatever the office. Yeah. Yeah.
00:49:33 - 00:49:42
I'm used to that. I'm dating myself here with, uh, those shows that I watch. Sanford and Son.
00:49:42 - 00:49:51
Yeah, exactly. Yeah. Right. I was watching, again, I date myself with this one, but I used to enjoy watching this with my dad.
00:49:51 - 00:49:59
It's on Peacock. I used to enjoy watching the original Night Court. I don't remember, I don't remember watching that.
00:49:59 - 00:50:14
Yeah. Yeah. That's way, way before your time or a little bit. Probably just a little bit. Yes. Anyway, how are, um, prices for eating out everywhere in Louisiana right now?
00:50:14 - 00:50:36
Awful. Like everywhere else, especially cause everything's a chain, right? So. Right. My wife and I were commenting last night. We went to this place called, uh, Junction 54 and she had two glasses of wine and I had one white claw along with some potato skins and she had a meal. And it was a hundred dollars.
00:50:36 - 00:50:38
Yeah. Ridiculous. Yeah.
00:50:39 - 00:50:53
I don't know how anybody that is raising like a family, like you got, like you're a family of four, right? Uh, I don't know how, how you do it. Like going out like four people. It's gotta be every time you go out 120, 130 bucks.
00:50:54 - 00:51:02
Not often. Or more. We don't go out often, but like we try to pick, you know, more economical places and even places like Olive Garden.
00:51:02 - 00:51:13
It's easy over a hundred dollars. Yeah. Right. Wasn't the Olive Garden pretty good about having kids plates or no? They do. They do. Yeah. Yeah.
00:51:13 - 00:51:26
Are you an Olive Garden fan? For the price, it's not too terrible in taste. Yeah. Yeah. Where do you guys like to go? Well, it kind of is dictated by like what the kids will eat. So Olive Garden, steakhouses, things like that.
00:51:27 - 00:51:36
Yeah. Yeah. Otherwise they get what a drive-thru McDonald's or cooking at home or something. Pretty much cooking at home. Yeah. Yeah. Yeah.
00:51:36 - 00:52:02
Anyway, I think we about hit our hour threshold here. I think we had another good episode. So as always, I appreciate everybody for listening. Our listener numbers were actually double what they were the previous week. So I actually submitted this podcast to five more podcast directories. So some people must've found this or that or AI bots are listening to all of it. Listen to all of us talk about AI.
00:52:03 - 00:52:10
So as always, thank you all for listening. DJ, anything else you want to plug or say? Nope. That's it.
00:52:10 - 00:52:17
We covered it. You're good. All right. Thank you everybody until next week for another insanely great podcast. See ya.