We Got Slopped in 2025

00:00:00:00 - 00:00:12:06
Unknown
This episode of the AI Inside podcast is sponsored by your 360. AI. Get 10% off through January 2026 by using code inside.

00:00:12:08 - 00:00:46:29
Unknown
Coming up next, Jeff Jarvis and I unpack Nvidia's surprising grok deal. Meta's $2 billion bet on Manus and the rising political pushback against AI in the US. Plus, we close out 2025 with a look at the year's biggest AI milestones and a hint of what's coming down the pipeline for 2026. All that more coming up on the AI inside podcast.

00:00:47:02 - 00:01:04:05
Unknown
Hello, everybody. Happy new year! Almost like in a couple of days. I'm one of your hosts of AI inside Jason Howell. This is the show where we take a look at the AI that is layered throughout the world of technology. Joined, as always by Jeff Jarvis. Good to see you, Jeff. Hello, boss. How are you doing? All right.

00:01:04:05 - 00:01:19:18
Unknown
We are squeezing in a quick episode, before the end of the year tomorrow. You know, at the time of this recording, tomorrow is New Year's Eve. And I was like, do we want to do a podcast on New Year's Eve? Turns out I'm doing another thing on New Year's Eve anyways, so it's not like I get the day off.

00:01:19:19 - 00:01:38:15
Unknown
Party boy. No. Not even. No, it's not that. But it is. You know, it's sometimes you just got to take the opportunities when they come and you think maybe people are going to watch. But last week's episode with Ben Gerstle, is, is is streaming on YouTube. So that was the Christmas Eve episode, as it turned out.

00:01:38:18 - 00:01:59:10
Unknown
Got to say, doing it, doing an awesome, amount of traffic and which I kind of had a feeling was going to be the case. Ben is a pretty in-demand kind of figure when it comes to the world of AI. So if you have not checked last week's episode, we obviously we did not have a news episode in the feed, but we did have our interview with Ben Gertz all.

00:01:59:12 - 00:02:19:24
Unknown
And, you know, it went really well. Ben's a fascinating character, fascinating individual in the world of technology and in the world of AI. Literally coined the term, artificial general intelligence 20 some odd years ago. So he's an important voice. And, that is evidenced by the fact that we've had a lot of people tuning into the interview, at least on YouTube.

00:02:19:25 - 00:02:39:01
Unknown
So, thanks again to the Singularity Net team for allowing us to chat with Ben Dressel. Yep. Yeah. Making it happen. And, I'm sure we're going to have lots more wonderful interviews coming up in the new year. So we've got some news today to talk to you about. If we would have had no news. Quiet week between nothing happening.

00:02:39:01 - 00:02:59:05
Unknown
But we got some big stories, actually. Hey, I mean, it just keeps on going. And. Yes, so we've got a couple of big stories, more on the like acquisition, acquire kind of side of things and a few other things. And then near the end of the episode, instead of doing our normal, like speed round thing that we've been doing, I figured we'd take a little time to just kind of do a little reflection.

00:02:59:05 - 00:03:21:06
Unknown
You know, it's the last episode of the year. Why not look at 2025 and I and not not at it, like a cohesive like this happened and then this. Oh, no. This happened. We're too lazy. Yeah, yeah. It's just too much to look back on. Geez. It's it's so much that's the problem. So I thought, why don't we just kind of pick out a few things that really stood out to us?

00:03:21:06 - 00:03:40:06
Unknown
So that's what we're going to close on. But before we get there, let's talk about some of the news that is right now. And I think this week, even though it has been a holiday week, some big news in the acquisition and the Aqua hire front, consolidation time, maybe, maybe they're trying to get this in before the end of the year.

00:03:40:06 - 00:04:06:13
Unknown
I don't know, but Nvidia has agreed to a non-exclusive deal to license grok. That's grok with a Q, by the way, that's important. Grok AI chip technology. They're also hiring away some of its top leadership, though grok is going to continue to operate. And, you know, like I've heard of grok. And again this is grok. Q not XS GR okay.

00:04:06:14 - 00:04:34:00
Unknown
So very important to note that but I'm not like intimately familiar with them other than kind of some of the research that I've done. They're known for, super fast Sram based AI inference chips. So this is this deal is very largely around inference. In AI, their chips is their systems. That means, you know, low latency, high predictability for performance for chat bots and for LMS that sort of stuff.

00:04:34:03 - 00:04:55:11
Unknown
And, like I said, it's all about inference. So it's targeting this really competitive inference market that's happening right now. To kind of bring down cost, increase speed and take advantage of of that. How familiar are you with grok? I wasn't at all. And I was I'm a bit of an official hero of what you need be Joneses.

00:04:55:14 - 00:05:14:15
Unknown
Yeah. And I've seen the he has short versions on YouTube. I subscribed and so I watched a 36 minute explanation of this deal. But Nate was, was very good, very clear about it and emphasizing exactly what you said. Jason, this is about inference, that is to say, the output of AI rather than the training of AI.

00:05:14:18 - 00:05:37:13
Unknown
Nvidia owns the market for training chips. And and there's a huge, shortage of what the versus what the industry demands right now. But this is on the other side. And they're betting Jensen Huang is betting that the models are trained, the models are commodified. The models are going to change. Yes. There'll be more training. Yes. But the real action is going to be.

00:05:37:13 - 00:06:12:22
Unknown
And hoping that this becomes a major business with major usage by corporations and individuals. That's inference. That's using the model to give you the output that you want. And what Nate explained about this is that the grok, chips have a fairly small amount of memory, but as part of the processor. And so it's extremely fast access, which makes it, faster and cheaper to then be able to, produce the inference output.

00:06:12:25 - 00:06:35:20
Unknown
And so he was very complimentary of, Jensen Wong, saying this was highly strategic, that rather than saying and I've seen this others have said this. Well, rather than Jensen was saying, well, we got to build this ourselves. He said, no, they're good. They're out there. And he went after them. He went after the unusual deal again, as you said.

00:06:35:22 - 00:06:56:06
Unknown
It's, the first reports were that he bought the company music at $20 billion. He would have bought the company. But it said that this wasn't a license. Company stays out there. I don't know what its ownership stake will be in. And grok, I didn't find that out. But then he got the top two executives or thereabouts from grok over to some major, right over to a giant Nvidia.

00:06:56:09 - 00:07:19:19
Unknown
Okay. And that's Jonathan Ross, who's the founder of grok, and then Sunny Madre, who's the president. So they move over. And what's so interesting to me from a dealmaking standpoint, just this whole, this whole point we find ourselves in right now because we keep hearing about the acquire without the acquisition strategy. And, you know, you've got Nvidia doing it.

00:07:19:19 - 00:07:45:01
Unknown
This isn't the first time Nvidia has done you've got Microsoft, meta, Amazon all employing this strategy. What is grok left with if their top founder like their founder and their president, the people who you know of arguably led the company to this major point. Suddenly they're hollowed out. What what is grok left with? Well, but again, Nvidia didn't buy it.

00:07:45:01 - 00:08:03:16
Unknown
Simply likes the company. So is it right? It's not. It's just right is it? If Nvidia has a stake. And again that's the part I don't know if they have a stake in the equity in grok, which I assume they do for that amount of money. It's in their interest for grok to keep succeeding. It's a very different kind of structure.

00:08:03:21 - 00:08:26:07
Unknown
Your journey is somewhat like this. This is not the same model, but they have interlocking. A corporate structure is where everybody owns a piece of everybody else. And they all kind of benefit if it happens. This is different from that. But I think it starts to build a new kind of structure here. Where they could do that and, and he gets the technology, he gets the view of it.

00:08:26:07 - 00:08:58:07
Unknown
He expands his business. The other thing I've read about this is that, Google with the TPUs, is somewhat, I would get this completely wrong, and I apologize for being out there. I just don't know enough. Google, Google Cloud and DeepMind models and Gemini, were a bit more like what grok brings, but Google is not selling their chips, at least not at scale to the world.

00:08:58:09 - 00:09:35:14
Unknown
Right? Yet. Yeah, right. Yeah, I think they will, but they're not yet. So Nvidia is and it's it rounds out Nvidia's, offerings. It is hosting in a way that becomes competitive with Google. It sets them up strategically. That's yeah. It makes them more integrated. And in in that sense kind of covering all of the different aspects. Now covering on a deeper level the inference aspect of their business, you know, so that they can compete with, you know, AMD and I guess Cerebro is another company that, you know, competes heavily in this space.

00:09:35:16 - 00:09:59:15
Unknown
Now Nvidia, you know, is really moving to kind of control maybe not own because they didn't buy grok outright but control internally within their systems. All of those points of the stack. Yeah. Right. And as we've talked about often everybody thinks of Nvidia as only a chipmaking company. But they're much more than that. Cuda is critical. And to keep people on the Cuda platform for Nvidia is critical because that also affects them.

00:09:59:15 - 00:10:26:02
Unknown
Two Nvidia chips and Nvidia hosting a whether that occurs at Nvidia, whether that occurs at AWS or anywhere else. So Jensen Huang is highly strategic and figures it out here. So I also read Cecille G. Tamura on Facebook, who finds a lot of interesting papers and explains some of these deals as well. And she puts it this way that in the competitive, landscape now.

00:10:26:03 - 00:10:52:20
Unknown
No, I just lost it, didn't I? There it is. Google owns layers, but not the broad industry lock in. AMD is a hardware challenger, but a weaker ecosystem. Gravity. OpenAI is a model leader, but a buyer of compute and not an owner of it. Meanwhile, she says Nvidia plus grok could unify deterministic real time inference, flexible training, and GPU throughput.

00:10:52:23 - 00:11:23:12
Unknown
DPU offloads don't ask me to explain it, to keep data flowing, CPU orchestration, and so it's really interesting just to understand the fuller strategic scope of these companies. Well and yeah, kind of like what we were saying dominating the platform as is it Cecile. Cecile. I think so. Yeah. Yeah. Cecile I wanted to say the first time I saw it, I thought is that.

00:11:23:19 - 00:11:45:10
Unknown
Oh I don't know, maybe it is could, could be either. But that's what Cecile says here. You know, it's not just competitive advantage. It's platform dominance. The whole the whole integrated platform, you know, soup to nuts, right? And, yeah, it's that's interesting. And by the way, you brought him up, Nate Jones. I'm such a fan. It would be cool to have Nate on.

00:11:45:16 - 00:12:10:03
Unknown
I've tried to get to all of them, and I haven't got any response. Yeah. He's great. I finally so it's. I just described John and he goes on these videos at length, but he's very good to explain. He also loves the tools and uses them in interesting ways. Yeah. Well, he's he's got a very I would say if I had to summarize my, my perspective on him, he has a very sober perspective on how to use these tools.

00:12:10:05 - 00:12:31:24
Unknown
That isn't over overblown. It it he does a great job of identifying and flagging, you know, the downsides of things, but also really smart ways to use these things. I've learned a lot about. Yeah, there was there was one. This is such a side tangent. But since we're talking about Nate Jones, I can, you know, go ahead and mention it.

00:12:31:24 - 00:12:50:20
Unknown
But I had set aside he had mentioned at one point Self-correction Systems. Did you did you catch his video on this? Of course I didn't. I didn't have access to a Substack before then. You gifted me the month and I really appreciate it because I'm super I'm super fan of his from what I've seen on YouTube. So I'm really going to dig into this.

00:12:50:22 - 00:13:22:23
Unknown
But the whole idea is, is that, you know, usually when you're using these LMS and you're using these models as we know they're going to often more times than not give you the thing that they think you want to see versus really questioning and being critical on their own output. And so his whole idea or that he shared, I don't know if he coined it, but what is chain of verification, which is this idea of requiring a verification loop inside of that same conversational turn.

00:13:23:01 - 00:13:44:23
Unknown
So essentially I want you to do all these things. Now, once you've done all these things, identify three different ways that your analysis, it might be incomplete. Once you have those three ways, site specific language that confirms or refutes your concern about those things, and then revise your findings based on all this stuff and based on your verification.

00:13:44:26 - 00:14:02:12
Unknown
And I've been using that for the past month, and that feels like a little cheat code, because I get such higher quality stuff out of these systems by doing that. So Nate super smart, he's got great suggestions and ideas on how to use these things. And, I highly recommend following his work. He's awesome. Yeah, he's a great explainer.

00:14:02:12 - 00:14:17:09
Unknown
And he's he's I would say that he's very much an enthusiast and that comes across. But he's not a fanboy. He's not crazy about the stuff. He's not going off into AGI clouds. He said, this is what just happened. And let me tell you what it can do. Let me tell you what I totally said. Let me explain how it works.

00:14:17:09 - 00:14:32:24
Unknown
So I find the baffle and he always has. I mean, fresh perspectives on everything. As you can see on his Substack, there's a post every single day, like literally every single day. And they are all accompanied with 20 to 30 minute videos. I mean, I don't know how this dude does it. Like, I look at this and I'm exhausted.

00:14:32:24 - 00:14:50:17
Unknown
I'm like, how do you keep up? I have to watch it all the cadence. No, but I'm but I have a massive amount of respect for the fact that he can do this because, boy, I can do this. That's just that's just insanity. Anyways. Right. Well, super. Well, if anybody out there knows Nate, tell them that we're.

00:14:50:19 - 00:15:18:05
Unknown
We are fanboys of his. Heck, yeah. Love to talk with Nate. I'd love to pick his brain on this show. Maybe we can make it happen. So there is that Nvidia and its aqua hire position. We thought that was enough. Enough news for the amusing week. Yeah. You thought that was it for consolidation time extravaganza. We have also happening this week meta buying madness, which is is a menace.

00:15:18:05 - 00:15:45:20
Unknown
Magnus I've always said, man, everything needs to come with a an a pronunciation guide. Yeah, totally. Or the parentheses with the phonetic pronunciation next to it. This is a very buzz, buzzy like it's been catching a lot of buzz this year. AI agent startup meta's deal to buy man reportedly worth around $2 billion. That's the valuation that Magnus was seeking for its next investment round, by the way.

00:15:45:20 - 00:16:12:07
Unknown
So meta just kind of said here it is and boom, we own you. Madness. A little background Singapore based startup, earlier this spring, I think, is when we first started hearing about this company and they were showing off their agents doing things like screening job candidates, planning vacations, analyzing portfolios. And, you know, a lot of people were very impressed by what it was able to do.

00:16:12:09 - 00:16:39:23
Unknown
So now it's a meta, property. Right. So this company actually started inside China, and when they got the announced the last investment from, benchmark, they moved to Singapore. And they're already been controversial because some people in Congress are complaining about, benchmark having invested in a China based company, but now they've cut their ties to China.

00:16:39:23 - 00:17:01:03
Unknown
I don't know what China thinks about that. And I don't know where the assets are and so on, so forth. But they're bought lock, stock and barrel by, by meta and the agent say are a couple things here. One is that that, it brings agents are the hot thing right now. So it brings agent expertise in two.

00:17:01:04 - 00:17:27:20
Unknown
It's a I think primarily B2B use. It's probably like corporate customers which could expand betters, revenue. Picture three. It's a revenue company now, so it brings in some revenue to, that is rather struggling AI business strategy. Yeah. They've got a lot, of lot of, recurring. What was it? $0 when it launched, obviously, to $100 million.

00:17:27:22 - 00:17:48:04
Unknown
Fastest startup. Oh that's right, that's what they said. Fastest startup to go from $0 to $100 million in the world, because only eight months since launch, they crossed $100 million in annual recurring revenue. That's pretty significant in a short amount of time. So it's another case, though, I think, where it I mean, even though we just praised, Nvidia.

00:17:48:04 - 00:18:10:01
Unknown
Right. Dead, for being strategically minded in going outside and acquiring by one means or another what you don't have, I think, to be a little unfair to Zuckerberg. It makes me ask whether he's still, is struggling for a strategy and is trying to acquire a strategy. You could argue because I'm would but I did friend video.

00:18:10:01 - 00:18:29:23
Unknown
Oh, he's very smart. He went out and this is hot. And he grabbed it and he used his money because he's got it to do that. And I think that's a perfectly legitimate argument. But I still don't know where Meta's overall strategy is headed these days. With Yang the corner leaving with his purchase with the prior purchase with, evidently some, service within the ranks.

00:18:30:00 - 00:18:53:11
Unknown
So it's gonna be really interesting to watch meta. In the next year. Don't count Zuckerberg out ever. He's got he's got enough, you know, money in the in the bankroll to definitely not count him out because again, he he can buy his way into things as needed. Yep. Exactly. It's one of the few companies that probably has, you know, relatively unlimited resources as far as that's concerned.

00:18:53:12 - 00:19:16:04
Unknown
If he's pulling resources away from the metaverse to I and believes this is the future. So, you know, we'll see where it goes. Yeah. I don't care, man. So keep running independently thinking maybe you already mentioned that Facebook, meta plans to integrate these agents into Facebook, into Instagram, WhatsApp on top of, of course, the meta AI, chat bot.

00:19:16:06 - 00:19:39:02
Unknown
And, yeah, it's kind of interesting in the, in the year 2025 for all that it is especially here in the US, a company like meta purchasing, you know, this this Chinese technology company has its ties to Tencent and Zen Fund and all these things which, like you said, they're shutting down its operations in China after the acquisition.

00:19:39:02 - 00:20:03:29
Unknown
But, interesting time. So meta stock first took a dive, as acquirers often do. But today it's up, 1 to 3%. So it's it's never last long, does it? Those two things happen. Yep. Just seems like that's the way the world right now. Yeah. Interesting. I mean, I have heard really, I have not, played around on a deep level with myself.

00:20:04:01 - 00:20:24:23
Unknown
I've, I've, you know, thrown a handful of things at it over the last handful of months here and there for whatever reason. But, I have heard really good things about it. So, yeah. Be curious to see, what meta does with this and to grab something away from China. That's okay. Yeah. No, I suppose so. So maybe that's why.

00:20:24:26 - 00:20:44:27
Unknown
Maybe that's why it's okay. You know, for, for, meta as a company, you know, a US company to, to buy China because, see, we're taking it from them, you know, in the, in the eyes of the U.S government. Anyways, if that makes any sense. Okay. So first of all, I want to thank our, our supporters on Patreon.

00:20:44:27 - 00:21:17:16
Unknown
Y'all are awesome. Patreon.com AI inside show. We deeply appreciate your contributions to the show. And, stick it with us. You know, we're coming up on two years here soon on the pod with the podcast, which I kind of can't believe. That's kind of crazy to me. How fast time flies. Big thanks to the patrons of the week, let's say our tech, Philip White and, my colleague, just a few of our amazing patrons who support us on a monthly basis through Patreon.com, slash, AI and sideshow.

00:21:17:17 - 00:21:34:18
Unknown
Thank you so much for doing that. And, few of you I have the, the the swag, the Google search swag. I think I announced it on last week's interview episode. Picked a couple of winners there. I got to hit the post office today. I've got all the swag all boxed up and ready to go, but you're going to get them.

00:21:34:18 - 00:21:41:21
Unknown
So you're going to get these shirts and this hat, whether you like them or not. We'll see. We'll see how you feel about that.

00:21:41:23 - 00:21:59:00
Unknown
Also want to take a quick break and thank the, sponsor of this episode. Oops. It's not Jeff Jarvis. Although thank you for being here, Jeff. You're awesome. Thank you to, the sponsor of this episode, your 360 Dodge.

00:21:59:01 - 00:22:23:03
Unknown
I just absolutely love doing this, this whole process with the folks at your 360 Jared Guralnik, of course, had joined me on the on the channel to take a look at the product, and now I'm going through the process, I think I think everyone, including you, Jeff, have, have given me my review, but I haven't gone through the portion where I check in on the review that's next in my process.

00:22:23:05 - 00:22:41:01
Unknown
So I'm going to check that out. Let's see what it's all about. I'm excited to to see what everybody said about me. But, here's what's really interesting here, and why this is so applicable. The lack of career development is the number one reason people quit a job where at the end of the year. So this is important time to think about this stuff.

00:22:41:08 - 00:23:06:18
Unknown
Most of us have never gotten feedback that was really just good enough to actually act on. And to improve our craft. Definitely true for me. Well, your 360 does something that was much more challenging even a few months ago because we have the rapid pace of development in AI. This uses a voice AI to conduct real conversations with you and your colleagues, 15 to 20 minutes each drilling in on specifics.

00:23:06:20 - 00:23:24:12
Unknown
The ways that you've done well, your growth areas, everything in between. And then it synthesizes all of that and walks you through the findings step by step in a coaching conversation, which I have as my next part of the process. And if you're a manager, it surfaces patterns across your team that you might never see in a standard survey.

00:23:24:12 - 00:23:48:19
Unknown
Really powerful stuff. A PM at Dropbox called it the most helpful, actionable career advice they've ever gotten, and you can check it out for yourself, too. You can start the year with real clarity at your 360. I use code inside. Through the month of January, you get 10% off. That's your 360 data AI code inside and you'll get 10% off through the month of January.

00:23:48:19 - 00:24:17:21
Unknown
And I got to check in with you, Jeff, just real quick, because I know you did the you did the process for me. How'd that go? So, yeah, as part of that part of working at, I was giving Jason a, a 360. I've never been crazy about three 60s, so. All right, so I go in, and it's a fascinating process because it's a it's a woman as the agent of choice, asked me these questions, and they're and they're clear and good questions, but a couple times I decided that I was going to be a jerk as I am.

00:24:17:29 - 00:24:32:07
Unknown
So I wanted to push the AI and see what it would do. And there was I forgot what the question was, but but she asked, and I kind of didn't answer it, and she came back. I think, two more times. Try to get it. I said, you've already asked me that question three times. And you want I'm sorry.

00:24:32:08 - 00:24:52:08
Unknown
You're right. She said she she John. And the somebody else. There's one other point and it were where I pressed I think I think you know she presses for for weaknesses and I always hate that question. If I had to, I'd say it. So I got copy again and she tried again. And it was absolutely fascinating to watch.

00:24:52:10 - 00:25:12:01
Unknown
So, then I got e-mail from Jared, said, are you. Well, your your conversation got flagged as one as, having complaints. Kind of. I was so embarrassed. You sponsored everything. And I said, no, no, no, no, I was just having fun. I just decided to see what it would do and push it. Yeah. And it did very well.

00:25:12:01 - 00:25:33:26
Unknown
It did. It did superbly. I would say that whether or not there was a buzzer. But the other thing that was fascinating to me about it, I said this to Jared, is that if I were doing this with a real person, I would be more constrained. This made me feel a little freer that there wasn't a factor of embarrassment or that kind of stuff.

00:25:33:26 - 00:25:49:09
Unknown
With this stuff, you just try to answer them saying the wrong thing, like like having an intention of what you're saying, but it coming out wrong. Whatever. Getting read wrong. Like, oh, I guess he's not that fun of Jason. If he didn't say he was the most brilliant man ever, right. That kind of stuff. And so it was oddly freeing.

00:25:49:12 - 00:26:10:21
Unknown
It also freed me to be a jerk. But I was a little reasonable lady. You've already actually got a stress test. Yeah. Yes, they did it, did I? It was, it was. And she needed to do what she did. I say she it, because you don't try to get an answer to this question, and I, I honestly didn't have one.

00:26:10:23 - 00:26:36:14
Unknown
But that's why I played it. And so hats off to Jared. Hats off to, your 362. I, if you do it, I think you're going to find it's more, a better experience than ever, having done a 360 before, because they can generally be a pain. This is a way to deepen it. And to that's not the whole pitch, obviously, but also to get the substance to make it useful for both the people giving it and the person receiving the feedback.

00:26:36:19 - 00:26:52:14
Unknown
And so I get it now and I think it's great. Yeah. Cool. Well, I can't wait to see. On the other side, I basically have an email this morning that's like, I think most of the people that I send it out to have completed it. So now I get the analysis part and then I get to see the hard truths.

00:26:52:16 - 00:27:08:19
Unknown
What am I good at and what do I need to work on? But I'm prepared for it. I'm all about, you know, I left out one thing, Jason. Oh, what? That one thing. Oh, yeah. Using gift as a verb. I'm sorry, man, I can't take that. I don't like gift as a verb. I would have added that in if I'd heard it today.

00:27:08:21 - 00:27:14:09
Unknown
So add that in your feedback then. Okay. All right.

00:27:14:11 - 00:27:36:21
Unknown
All right. Sounds good. You're 360. I code inside 10% off through January. Thank you, Jeff, for that. And, thank you to the folks at your 360 for sponsoring this episode of the. I work. Much appreciate that. All right. We're going to take a super quick break, and we're going to come back and we're going to talk about anti-air, or at least a reaction in the opposite direction to AI that we're seeing right now.

00:27:36:21 - 00:27:42:18
Unknown
A few articles. Very interesting, that's coming up here in a moment.

00:27:42:21 - 00:28:13:10
Unknown
All right. Let's let's take a look at this anti-air movement. As the author at the New York Times anyways wrote, this is Michelle Goldberg, writing for the New York Times. The article called an anti AI movement is coming. Which party will lead it? And, you know, there's a few articles that we can talk about here, but she argues in this article that there is a widespread public souring on AI that is happening.

00:28:13:12 - 00:28:42:24
Unknown
And I think that, you know, that's that's certainly in some it's true in some corners. That's true. Right? Yeah, exactly. She says driving this is a sense that it's less a miracle technology and more an erosion to our to work, education, mental health, privacy, perception of reality, all of the things she points out that, many of the people that are creating and promoting the technology have also acknowledged that it has dangers and flaws, but they continue to do it anyway.

00:28:42:24 - 00:29:09:04
Unknown
Something, you know, we've we've definitely talked about, on the show. And I think her broader point is looking at how the AI divide has an interesting impact, like on intersecting, at least here in the United States, the US political lines, because you've got like, you've got like President Donald Trump and some Democratic leaders who are really kind of supporting and courting the AI companies, lessening regulation, that sort of stuff.

00:29:09:04 - 00:29:43:07
Unknown
You've got Republican leaders and independents like Bernie Sanders kind of together in some capacity to push for more regulation moratoriums on the technology, that sort of stuff. And it leaves voters kind of in a strange position, I suppose, where they might support a candidate deeply on a number of things, but then really feel strongly, you know, anti or for their stance on AI and how it's just kind of playing with the, the political lines in the US to a degree.

00:29:43:09 - 00:30:03:28
Unknown
Yeah. So, so Politico had a very similar story. It's as if they'd read each other. Yeah. I don't really hate AI. Which party will benefit is their their take. And as you mentioned, Bernie Sanders, all Americans. Yeah. And then Bernie Sanders is now coming out and arguing for, a moratorium on development, which I think is ridiculous.

00:30:04:00 - 00:30:34:00
Unknown
And so it's fascinating to watch this, but if we look at it in the context of history, it's really easy to see how we've been here before. You go back in the 90s, section 230 was a bipartisan piece of legislation from Republican and Democrat that defended the future of the internet and is, to many people, including me, a critical piece of legislation, that you have others from both parties in recent years who were smashing section 230, arguing that it should be repealed.

00:30:34:00 - 00:30:59:19
Unknown
God help us if it were, and not really understanding what it is, but it's just an anti tech. You. And then I just finished reading before the show, a book by the famous, sociologist Raymond Williams in 1975 about television. So about as far away from the moment when television hit the majority of of homes in America as we are today from the turn of the century and, the web taking over that.

00:30:59:19 - 00:31:45:24
Unknown
So it's really. Oh, okay. Wow. It's a 20, 30 year old, technologies. And then, television was seen in great measure, not as a cultural institution, but as a technology. And then go back another 50 years and radio was seen not as a cultural institution, but as a technology. And the reaction of many in the public and of, the cultural institutions, critics, academics, politicians and so on is to a technology rather than, I think, to what inevitably follows once people start using it, making it their own, and it becomes demystified as a technology.

00:31:46:01 - 00:32:06:25
Unknown
So we're in, I think, a standard phase, where people think that, oh, no, it's technology that's going to it's going to eliminate all jobs. It is going to destroy mankind. It's whatever, whatever those arguments are. And I think that that becomes, specious in the long run. But damage can be done in the short run with bad legislation and regulation.

00:32:06:28 - 00:32:45:02
Unknown
So this I mean, the real Goldberg is just assuming that, you know, an AI anti-war movement is coming. It's going to be there. Politico declares Americans hate AI. That's ludicrous. But that's where we are in the the kind of digital Puritanism of our media structure. I mean, there you know, I can't remember which of these articles or maybe it was an entirely different one, but I, you know, saw Gallup polling that shows 80% of U.S. adults think there should be governmental regulation of AI and a and a like an actual slowing of its development.

00:32:45:02 - 00:33:22:03
Unknown
There was a Pew study that shows that, 17% of U.S. adults see I having a positive influence on the country over the course of the next 20 years. So then, you know, I suppose the alternative, you know, the outside of that is 100 -17 is the not positive, I guess. Right? So, I mean, there's probably some truth to the fact that we might be looking, at least to some degree at the next, you know, election cycle, having, having AI as, as a pretty involved kind of element to how some of these decisions are made or.

00:33:22:06 - 00:33:45:20
Unknown
Yeah, I wonder whether where people would really put that on their list of things that matter. Yeah, that's what I'm curious about. You totally. I agree, I mean, I think it's it's the economy stupid or it's stupid economy, as the case may be. I think the immigration stuff is going to matter. Tariffs matter. People's health matters with what's happening with Kennedy.

00:33:45:22 - 00:34:04:21
Unknown
My politics are showing. Sorry about that. And so I think those issues are front of mind. I think media are I think this is a year end phenomenon. Oh, I got to write something for the last of the column. Yeah. No, I mean, these these articles are everywhere. Yeah, exactly. And so I want to declare this war against AI and yeah, it could happen.

00:34:04:24 - 00:34:30:00
Unknown
But I don't know that it's politically it's going to be, a high level issue for people. Yeah. Because because the so-called threat is as amorphous as the so-called promise of AGI. Well, yeah. So if, if I can, if you'll indulge me, I have a short little kind of related personal journey here that I think ties into this.

00:34:30:03 - 00:34:51:06
Unknown
I saw mine a minute ago. So it's your turn now. So okay. So the last, the last month or so I've been testing out and kind of strategizing around a part of my business, being around podcast consulting, some sort of podcast education, be it consulting with people who want to start a podcast. And hey, I've been doing it for 20 years.

00:34:51:06 - 00:35:15:06
Unknown
Let's talk. And, you know, maybe I can help you. So I've, I've been talking with people just for free, just for fun. You know, I kind of put it out there like, hey, if you got a podcast, reach out, whatever. And it's been going really well. It's been a ton of fun. And so as part of this have been thinking like, okay, we'll go in in 2026, what would it be like if part of my work was, you know, that as an arm of my, my business, you know, a consulting, podcast, consulting arm.

00:35:15:08 - 00:35:33:15
Unknown
And so I thought, what I need is I need a landing page that that is something that I can point people to, to, you know, like a form that says, okay, what is what is your idea? What is the challenge that you're facing? Here's a few pieces of information here. You know, like who's your audience, who is definitely not your audience, all this kind of stuff.

00:35:33:17 - 00:35:54:05
Unknown
And the idea with that form, of course, is fill out the form, then, you know, now I have the information to know that I can help you. And they get an email in their inbox that says, hey, you know, here's here's what you sent me. Here's what I think about it. Let's, let's fire up a consulting conversation. I'm making this up as I go, by the way.

00:35:54:07 - 00:36:18:24
Unknown
But I thought that sounded like a great idea. So I spent two days of the holiday break building this out with a working with, like, a form filling site. I can't remember if it was Typeform or something else. Feeding that into Zapier, having Zapier take the the form and put it through like a Gemini instance and kind of understand it a little more, and then craft an email and send that over to Gmail and also put it into spreadsheets.

00:36:18:24 - 00:36:36:26
Unknown
And so I was like, I challenged, happy ending here. It no, no, I think it heads to an interesting ending. Okay. Nothing, nothing dramatic or horrible comes at the end of this. But I finally, after a couple of days, built this tool and I was proud of it. I was like, hey, you know, I set out to do this.

00:36:36:26 - 00:36:54:09
Unknown
Great. I'm going to send out a link to, you know, a handful people and just have them stress test it. Like, can you just go to this thing and fill it out? And what I found kind of surprised me because I'd say the reaction of at least half the people, mind you, these are not all people that are a technology like in our tech circle.

00:36:54:09 - 00:37:14:27
Unknown
Some are, but some are outside of it. And I'd say definitely on the people who are outside of it. Largely, I got a very not angry, but a visceral reaction to the fact that there was AI involved at all that that like, oh yeah, your form was great. But then I got the email, and the email felt very inauthentic.

00:37:14:27 - 00:37:49:22
Unknown
It you know, they're like, we know your brand, we follow you, we're a fans. And the email doesn't doesn't sound like you. The you know, the I made certain assumptions that collided with my personal life and in coincidental but alarming ways. And it was just an interesting thing that I bring up now, only to say that, like, I think there is this kind of, reaction in the in the opposite direction that I'm noticing with people in my life that aren't in technology, that archive deep in this stuff.

00:37:49:23 - 00:38:11:00
Unknown
Right. Where, where people do, you know, it's totally anecdotal, I realize, but people who don't follow this stuff super closely are starting. Some of them are starting to feel kind of like an immediate negative reaction to AI as a thing. And that could be because of news stories like these. That could be for any number of reasons. But it's there.

00:38:11:04 - 00:38:29:29
Unknown
And I really wonder with that being there and kind of the reaction that I saw from this thing, what does that mean when we head into, you know, the next couple of years where the these tools build out even further, where government makes the decisions that it's going to make for right or for wrong to either bolster it or not.

00:38:29:29 - 00:38:47:08
Unknown
I don't know, I'm very curious about this. I'm very curious to see what kind of impact it has yet. I wonder how many of those same people who were suspicious of. I would say the same thing about Google and Meta and Amazon. Yeah. Is it an anti, tech company? Is it Anti-big company? It's like capitalism. I'm not trying to read too much into it, but I think there's a lot of that that comes in.

00:38:47:08 - 00:39:06:22
Unknown
Or is it specifically AI that has cooties? I'd be very interested to be very interested to do a poll of them, almost go back to them and say, okay, I hear you, I got it. I don't know how you arrange that, that question, but it'd be interesting to interview them to see where their hostility really lies.

00:39:06:22 - 00:39:26:11
Unknown
And I think it'll say a lot. Yeah, yeah, it's it's a curiosity that I have, though. Yeah. And, Daniel in, in chat said, did you feed the AI some examples if you're writing. So it could possibly write in your own voice. Yeah, absolutely. Did I think, I think though there are just certain tells AI has tells.

00:39:26:14 - 00:39:44:09
Unknown
And the more we use these things, no matter how good we get them, they're still there, whether we recognize them or not. They give us a certain feeling. You know? And I don't know, it's just interesting. Now it has me rethinking it. Like, maybe I don't want this thing to be too smart, right? Right. It's because it's you they trust.

00:39:44:12 - 00:40:18:12
Unknown
Yeah. So I just got caught by slop. Affiliated with the Alan Alda Center for Communicating Science at Stony Brook, really wonderful institution. And I've through that. I've had the honor of being Alan Alda and speaking with them a couple times. And just an amazing, amazing man. And I'm going through my Facebook. There's a journalist, a journalism educator I really respect to put the story up of how Alan Alda helped one of the cast members of Mash in the day, and so I shared it with my boss at Stony Brook, because she's good friends with Alan Alda.

00:40:18:12 - 00:40:36:23
Unknown
And she said, oh, what a wonderful story. And about an hour later, I had to come back and say, I got slapped. Somebody showed how it was slop and when you think about it, in hindsight, it was too perfect. It was too this. It was too that I. Yep, yep. Yeah. Got slapped, I got soft I love it I love that.

00:40:36:25 - 00:40:55:24
Unknown
But here's a better story. So I'm not going to say what it's about because I want to keep this quiet. And you may laugh at me for this, but I have an idea for a screenplay, okay? And I have no idea how to write a treatment. No idea whatsoever. And it's probably insane, but it was it same a holiday week?

00:40:55:24 - 00:41:15:27
Unknown
Oh, what the hell. I'll spend a day and a half doing this. And, I started by writing about a four page just narrative of it, and then I fed it to notebook, and I said, can you help me write a treatment? And it has all this for about anything. I've called that two at three to this the that there's a way to do it right.

00:41:16:00 - 00:41:43:18
Unknown
And I tried the same thing with GPT and Gemini and anthropic I think. And the formatting was the same in all of them. I think what they did was pretty much crap. It was still my story. I knew what the story was, so I wrote it, but I could have read, anywhere online. I'm sure I could find how to write a treatment, but then it's antiseptic and other by using my own material and making me critical of all I can do better than that.

00:41:43:18 - 00:42:03:21
Unknown
Well, then. Okay, Jarvis, do better than that. Okay? I will write. It was more useful as an educational exercise for me. Because it was, it was, it was relevant to me. It I understood what was wrong with it. I could go with it. And it was really kind of it was my first best collaborative effort with AI.

00:42:03:23 - 00:42:23:15
Unknown
So the idea was me. I threw out everything that it said, but I learned from it nonetheless. Was helpful. Yeah. And did it did it lead you to make different choices in what you chose to do from a creative standpoint afterwards? Maybe, you know, maybe, maybe it leads you to to oh, I hadn't thought about that particular thing or whatever.

00:42:23:15 - 00:42:42:06
Unknown
So maybe it influences you in ways other than, hey, I wrote your treatment for you. Do you like it? You know, the problem I have with it. So I wrote to my agent. I sent it to my agent. I haven't heard back from her. The the subject line of the email was, you'll think me insane. Dot dot dot.

00:42:42:08 - 00:42:59:16
Unknown
So I have no idea what she'll say. What bothers me about the I was the sycophancy. Wow. What a great story, Jeff. This is really filled. Good. Right. Well, I don't know whether that's true. It's friggin machine now. Right now. And I don't want to I don't want to take any of that in. Yeah.

00:42:59:18 - 00:43:15:05
Unknown
But it's fascinating to work with this stuff. And again, I wrote a syllabus for Stony Brook, for a course that I created, creativity, which I didn't teach to, to great people, did teach it this term. And it filled up a meeting with 300 students and next term, another person I really respect there is teaching is 500 students.

00:43:15:07 - 00:43:40:09
Unknown
And and yeah, there may be people out here who hate AI or want to growl about it or want to give it up or all that, but it ain't going away, folks. Yeah. Yeah, it's super interesting. Like, gosh, the show could be about these, these kind of stories because I think I think you and I having these experiences is indicative of what a lot of people are experiencing as we work with these things more.

00:43:40:09 - 00:43:55:04
Unknown
I mean, my wife is is working more with these. I got her I got her new laptop for Christmas. I was very proud because anytime I had to use her, her laptop for her business, I wanted, as I told my daughter, I just want to jump off the roof. It's so miserable and frustrating. Everything takes 1,000,000 hours just to do so.

00:43:55:04 - 00:44:10:18
Unknown
Got her a new laptop because I've been also binge kind of showing her what I've been learning about working with these systems for her business and everything. And she told me the other day, you know, she was sitting there, she was like, you know, these tools are fine and good and like, it's it's it's interesting and magical to see what they can do.

00:44:10:18 - 00:44:30:04
Unknown
Sometimes she's like, but the more I work with them, the more I realize how imperfect they are. And I was like, yeah, spot on. Like, yes, they seem so magical. And so, oh my goodness, it can do everything for me. Now when you first start, the more you get accustom and work with them, the more you really push up against the obvious limitations.

00:44:30:09 - 00:44:55:03
Unknown
Unless you fall down that that sycophantic root of yes. Oh, I am really great. I this this is amazing. This is the perfect way to phrase things or whatever. And then, you know, and and I think I kind of faced that a little bit with my tool and thinking like, oh, I built this great tool, but not thinking about like, what is the what is the message that I'm conveying on the other end when they get that final email?

00:44:55:06 - 00:45:24:24
Unknown
And it's definitely not authenticity, which is a core value of mine. And so okay, then I have to scale back and I have to think about this differently because that doesn't work for me. Yeah, yeah. Interesting stuff. I love that tangent. That was fun. Before we get to the archive showdown, just real quick, because everybody was talking about this, this story, OpenAI hiring a head of preparedness, a very highly compensated role.

00:45:24:26 - 00:45:53:19
Unknown
If anyone wants to put themselves up to, this role, 555 K thousand plus equity and, basically the whole idea here is this unfortunate soul has to monitor and contain risks, from the companies, your models, as they increase in capability and everything they have to address potential harms to mental health, autonomous cyber attacks, misuse and biology, misuse of weaponry, all of these things.

00:45:53:22 - 00:46:16:17
Unknown
And, even Altman admits, quote, this will be a stressful job and you'll jump into the deep end pretty much immediately. So, Jeff, are you up for the challenge? Because I'd walk in and say, this job is B.S., there's no AGI, there's no end in humanity. I'm going to sit back and take my paycheck. You're going to hang out on the roof all day and.

00:46:16:19 - 00:46:39:00
Unknown
Yep, yep. Eat your lunch. Free lunch. Probably. Probably. Oh, yeah. Probably a pretty good free lunch, actually. Yeah, I bet. Yeah. So anyways, I thought that that's interesting. Definitely. A lot of people thought that interesting. Do they think it's interesting because it's highly compensated or because it's OpenAI kind of. It kind of. They don't. But I try to argue we're going to kill the world.

00:46:39:00 - 00:46:59:15
Unknown
So you're you're the person who's going to save the world from us, which is so weird. Yeah, that's that's marketing. Are they are they going down the anthropic route with this? Oh yeah. It's like, well, don't forget that anthropic people came from OpenAI. And so they're similar. That's true. Cut from, cut from the same cloth as they all seem to be.

00:46:59:17 - 00:47:23:07
Unknown
All right. Well, why dance around it any further without some music to dance by? As you might, it's time for Geoff's archive showdown. Geoff. So I got you down. Down and out. Okay. What do you got? Three quick papers. Here for my, two weeks worth of trolling. Archived on the preprint server. The first one interested me is sell it before you make it.

00:47:23:07 - 00:47:51:27
Unknown
Revolutionizing e-commerce with personalized AI generated items. Importantly, this comes from China. So this is not going to be anything unusual to somebody who knows. Kickstarter. But what they're doing is they're using AI to design, products without having to prototype them. And then they can put them up and they can see, do people want this? And if they want it, then we'll make an offer for it, and then we'll take the orders and then we'll make it.

00:47:52:00 - 00:48:21:24
Unknown
So it's it's an extension of the Kickstarter model where you use your customer's capital to build the thing. But in a world of, high turnaround fashion, this will be even more high turnaround. And it's not hard to see. I think this is not really part of the paper, but then I could see it. But why does the tool need to be in the control of just the company if if I designed, a jacket and I convinced enough people to say, do you like my jacket?

00:48:21:27 - 00:48:45:00
Unknown
And, what if, they could discover that, Yeah, they'll make it for me and they'll sell it to me and they'll give me a cut of it because I designed it to. What? Can I use AI to do that? Or can I get a single, purely personalized item because I can do it. So it's just AI as general is now moving in to this, sequence of product development.

00:48:45:00 - 00:49:02:08
Unknown
And I've long argued, AI argued, I think it's interesting that you've got to move the customer up much earlier in the design chain, rather than designing the whole product in secret and saying, here, don't you love it? And we spend a lot of money to prove that you love it. The more that it comes from the need and desire of people themselves, the better.

00:49:02:10 - 00:49:29:22
Unknown
So then you know that what that has me thinking is, we've talked, in the past about, I don't know if it's a possible future. I don't know if it's an inevitable future, but the future where? And, you know, maybe things exist. Like maybe, maybe movies exist. But there's also like, an, tangent where people, the systems are good enough to where I have an idea and I say, you know, the movie I want to watch right now is a movie about this guy that blah, blah, blah.

00:49:29:22 - 00:49:46:23
Unknown
I spell out the the basic plot line and bloop. It creates the movie for me, and it's good enough that it's entertaining and blah blah blah. I wonder if that could happen with products too. Or it's like, you know what? What I really want is a sweatshirt that does this. Just make it and produce it and send it and that I could I could see that happening.

00:49:46:23 - 00:50:02:14
Unknown
Like I could absolutely see that happening with systems like this. So, Google announced some things that they that they changed. And one of them is, that they said to me to see if this if I could get this to happen,

00:50:02:16 - 00:50:25:19
Unknown
Tell me a story about an entrepreneur who starts a company and becomes rich and rich in Seattle. Elias Thorne didn't feel like a future billionaire. He felt like a man who was tired of losing his keys. Elias was a software engineer with a frustrating trait. He was obsessed with micro inefficiencies. He noticed that people spent an average six months.

00:50:25:19 - 00:50:50:00
Unknown
Okay, that's enough. Yeah, but one of the things is Google. Gemini is now saying, we'll tell you a story. Yeah, exactly what you just said. Tell the story you want to be told, Yeah. And I mean, you know, is it a good story that that remains to be seen? Probably not. But I mean, long enough timeline, you know, enough development, enough training around what is good versus not.

00:50:50:00 - 00:51:12:02
Unknown
I guess that's the real challenge. But yeah, I could see that. And so, you know, are we are we looking at a future where it's like that. The clothes that you wear potentially are the clothes that you just had an idea. You wanted and you didn't have to, like, spend 1,000,000 hours trying to find the thing that matches you just say, I need a I need a black turtleneck that fits a tall size, blah, blah, blah, and has this logo on the back, or I don't, I don't know what it looks like.

00:51:12:02 - 00:51:36:23
Unknown
And then the AI just goes, Boop, here you go. Yes. Let it out. Yeah. Six hours shipping. Second, second paper. Yes. Yes. Probably. Yeah. Instant manufacturing. The two houses down from you. Right. The factory drives up to your driveway and drops it off. Second is fascinating, especially what you talked about with your wife's attitude toward AI. What human course interactions may teach us about effective human AI?

00:51:36:25 - 00:52:09:03
Unknown
Interactions. Okay, this from Mohamed Hussein Johari and Stanley Alt at, Uncg Chapel Hill. And their argument is simple and straightforward, but but but charming for what it is saying that that the way that a writer and I'm not a horse rider are you do you ride horses? No, I no, I don't know, but I mean, I've got friends who are I live in horse country, and there's friends who are, but there's a certain magic that occurs of communication that the horse is the one doing what you're doing.

00:52:09:03 - 00:52:27:15
Unknown
But they they have a shared and harmonious decision making. The horse is deciding you're deciding you're coming together to do this stuff. And so it's an argument for a model around what you think of as AI, which I think is good. It's not that it takes over the world, it's that we work in harmony in tandem. Yeah. Yes, yes.

00:52:27:18 - 00:52:55:12
Unknown
And then the third paper from folks at, Institut Polytechnique in Paris and, a cool pillow technique, in Paris, fake parts, a new family of AI generated deepfakes. What's interesting about this is this is not a full deepfakes. This is the discovery of the ability of AI to fake just part of an image or part of the video.

00:52:55:14 - 00:53:19:22
Unknown
So imagine if you would, you can change the face, right? We know it could do that in photos. So the next phase of deepfakes, it'll be harder to tell because 95% of the photo is real. Wow. Okay, this part is faked. So. Okay, that's that's interesting. There's probably a term for this that I'm just not aware of.

00:53:19:22 - 00:53:44:21
Unknown
But when I think of, like nano banana, let's say, and nano banana has gotten really good at giving it a source image and saying this, but here. Right, or whatever, you know, and retaining the qualities or this image, but change my the hair to read my understanding about it is that it's not actually only changing the hair to read, it's doing a whole pass over the thing.

00:53:44:21 - 00:54:09:04
Unknown
It's just getting really good at replicating the original thing. I wonder how that applies to here, because in that case, is it really possible with current systems to only change a portion of the thing and not the entire thing? Just make it adhere really closely? If that means they list six? No, eight, possible sort of, which I don't understand, face swapping inpainting, which we've seen that, you know, you fill in part of the.

00:54:09:05 - 00:54:27:19
Unknown
Yeah, it was the interpolation. You extend something out painting, you extend the image, style change. Those are the six ways that they to sort of define it can be done. Yeah. And they're, you know, they're looking at it in a way that I can't fully understand because these papers are highly technical. But yeah, interesting trend are coming.

00:54:27:21 - 00:54:49:14
Unknown
Yeah. And I guess yeah, that's true. That that kind of that kind of answers my question. I mean, especially when you talk about face swapping. I mean, this is a technology that we've had for a number of years now. AI has inherently it's made it higher quality and and more effective. And in that application, we're not seeing the system completely generating the entire image.

00:54:49:14 - 00:55:07:14
Unknown
And just swapping the face. So that would that would seem to be the answer. Yeah. Is that yes, we do have the the ability to do this. So interesting. I mean that is that does make, you know, detecting a lot more difficult when 95% of the images is fine. And maybe it's that one really important smaller part.

00:55:07:14 - 00:55:27:09
Unknown
And all you want to change is whether it was a $1 tip or $1,000 bribe, you know? Yeah, it's all you're doing is changing one little aspect of the image. Yeah. Interesting stuff. All right. Well, this show is fascinating me to no end. Hopefully it is to you, too. And if you like it, leave us a review.

00:55:27:12 - 00:55:44:04
Unknown
I'll just say that wherever you can leave a review or comment on the videos or whatever, interact with the with the AI inside, wherever you happen to be watching or listening. We really appreciate it. Going to take a super quick break. Gonna come back with a few minutes to take a look at the year that was 2025, and maybe I'll look forward to 2026.

00:55:44:04 - 00:55:46:04
Unknown
That's coming up in a moment.

00:55:46:07 - 00:56:08:28
Unknown
All right. 2025 and review. And I think this is inspired. You would you would put in a link to Andre parties post 2020 white volume review. And this is for him. This is extremely short. Extremely short. Yes. Okay. So okay, I wonder I wonder why. Hey, you know what? We all have a lot of things going on right now, so it's okay.

00:56:08:28 - 00:56:34:28
Unknown
He's a little busy. He's a little busy. But. Yeah. So so this is interesting. So some of this stuff, I had to kind of like, noodle on and think, think about. I don't know how deeply we want to go into this, but, reinforcement learning from verifiable rewards. So this is basically like a training a step of training where models are learning their reasoning, in a, in a more enhanced way because they're chasing rewards that can get checked automatically.

00:56:35:00 - 00:56:57:14
Unknown
Is, I guess, one of his reflections from the year. He's he's probably I mean, obviously clearly he's looking at this stuff on a deeper technical level than probably I do. I, I'm not going to speak for you. No. Show jagged intelligence. At first I was like, what the heck does that? But it actually makes a lot of sense and kind of ties in with what we were looking at.

00:56:57:20 - 00:57:16:01
Unknown
We're talking about a little bit earlier. I can look really brilliant when you're talking about a few narrow areas, but then it falls apart on the tasks that, that surround that, tasks that seem like they should be very easy. And this is still a challenge with AI as far as what we're seeing. Any I mean, there's there's a few others here.

00:57:16:01 - 00:57:45:17
Unknown
Any stand out for you in particular, but a cursor is the hot thing that people are going for. Yeah. This year, I think, my son is using it as a developer, and I think it's an impressive people alongside clawed code is really been the winner of the year when it comes to coders alongside vibe coding. So in the coding world, you know, if, if these early stages of generative AI had stuck to coding, I think it would have absolutely blown the minds of a small but powerful part of the country.

00:57:45:20 - 00:58:18:26
Unknown
But it wouldn't have been the huge phenomenon of business that that it has been. Yeah, yeah, that's where is most successful I think. Yeah. And then it calls out nano banana and light guy which I, you know I yes nano banana is really impressive as we've talked about Gemini and some of their integrated Google's integrated kind of graphical hooks that are coming into the model are is getting really impressive and really good and kind of making it a more friendly, familiar kind of, way to do this sort of work.

00:58:18:26 - 00:58:37:15
Unknown
And so that's going to, I think, over time, broaden out and, bring more people in. So, yeah, kind of a kind of a good walk through. What are a few things that kind of stood out for you, for this year? I mean, obviously there are a bazillion things we talked about, but I'm curious to know. So, each of us has three things that we'll talk about from the year.

00:58:37:18 - 00:58:58:04
Unknown
My first is the split between limb believers and role model believers. And I think that we've talked about that a lot of the show. And, you know, Google is still going strong, in the, limb world. And they believe that it's going to and I went to this event with a DeepMind person and Yann LeCun, they ended up in a debate.

00:58:58:07 - 00:59:24:22
Unknown
So interestingly, there's a video of Yann online, which he says there's no such thing as general intelligence. Human intelligence is super specialized for the physical world. This is playing into exactly what he's doing in the future. He's working on, role models faithfully, you know, role models. Jensen Wong has talked about role models. So in comes Demis Hassabis, who then says Yann is just plain incorrect here.

00:59:25:00 - 00:59:46:17
Unknown
He's confusing general intelligence with universal intelligence brains with this exquisite and complex phenomenon that can take on general knowledge. And that's the argument on that side is that we've seen the performance so far. If we just scale it, it'll do anything. And then Yann comes back and says, I think there's a disagreement, largely one of vocabulary, that he objects to this idea of general intelligence.

00:59:46:17 - 01:00:08:01
Unknown
And he says that we are all specialized in our intelligence. There are certain things we can do really well. We can figure out the physics of a ball coming our way. But as he said in the video, we're crappy. Generally at chess, the machine can be better. At chess, we are specialized in our knowledge, so this is the essence of an argument, that's going on in mathematical terms, and it's still terms going on.

01:00:08:01 - 01:00:29:23
Unknown
So that's one. Two is the second thing for the year, I think my conclusion is Google's a new Google. And with a lot of people, big year for Google. Yes. People were declaring Google dead. OpenAI has taken over. Search is dead. No. Google's done a superb job and I think it's it's on top. And interestingly, I just threw this at the last minute.

01:00:29:25 - 01:00:56:07
Unknown
You'll remember that when Google put up the, this the AI as part of the search results. And somebody from Business Insider went in and got it to say, how do you get to keep the the cheese on the pizza? And Google said, you mix and glue, which everybody had to make fun of and indeed and the and the topless, Business Insider in fact made pizza with glue and wrote 0840, anything for a story or an entity, a story.

01:00:56:09 - 01:01:21:12
Unknown
But now big credit. She came back with the year end and ate crow instead and said that Google I o reviews had the last laugh this year that they've really been good. I guess Google has the last laugh now. I love that, they do indeed. And then she ate crow without glue. Yeah. My third is U.S. versus China, I think this year, because of deep secret.

01:01:21:14 - 01:01:40:26
Unknown
I think that's a huge part of the story. And it's going to have policies, political, strategic business applications going forward. It want to 100%, 100%. Yeah. And by the way, like, I don't know, there's probably plenty more. But when I was thinking about it I was like I was like, yeah, here's a couple of things that come to mind.

01:01:40:28 - 01:02:03:13
Unknown
And kind of what I was talking about earlier, just a general kind of thing that has me really interested right now. Is this this, like, confused state that I know people as more and more people are interacting themselves with AI, that I think a lot of people are feeling right now is, you know, both techies and non-tech AI's seem to, you know, they're using the tools even though they have they hate to admit it.

01:02:03:15 - 01:02:20:15
Unknown
There's like a conflicted relationship that I think a lot of people have a love hate relationship kind of confusion around around these tools because. Yeah. And I and I and the only reason that this is really top of mind is kind of like the story that I told earlier and some of the conversations I've had with people outside of tech circles.

01:02:20:17 - 01:02:41:25
Unknown
Lots of people are using these tools, and some people are being really, I don't know, some people are using them, but then don't want other people to know that they're using them. It's like they're ashamed to be using them. It's a weird time for this technology, you know, and for people to know, is it okay, is it okay that I use this technology even though everybody's using it and everybody's pretending like they aren't?

01:02:41:25 - 01:03:09:07
Unknown
So I don't know. I think that that's been, kind of an interesting thing to watch. Just the fact that, like, this was a big year for agents and yet, man, do they need a lot of work like. And I guess agents is a broad term. But when I think of, you know, from a consumer perspective, which is largely my own perspective, any of the times that I've really kind of leaned in to agents to test out something or try something or, oh, wow, that capability is here.

01:03:09:07 - 01:03:35:12
Unknown
We can finally do that. Rarely am I ever presented with, oh, I did it perfectly. Yay! I'm going to do that again next time, you know? But it's it's kind of the it's still early days for agents. As much as this seemed like the year of agents and again, I know that's a loose terminology and there are a lot of ways in which, you know, people have become really good at programing agents on a deeper level and getting them to do cool things.

01:03:35:15 - 01:03:57:21
Unknown
Yes, that's possible, but I think by and large, they need a lot more work before the masses, before more people gain their trust and actually decide, like, okay, yeah, I'm going to rely on this for buying or whatever the case may be. And then, did we even talk about the AI for companionship? I think maybe that was a story that was in there earlier that maybe got cut.

01:03:57:21 - 01:04:26:18
Unknown
But just as a general theme, this has been a really interesting year when it comes to the chat bots and their development and how that is working its way into general users from a companionship perspective. Lots of resistance. Notable adoption, but plenty of resistance. And I think there's little acknowledgment in that as we talk about little acknowledgment around the reasons that are driving people to do that in the first place.

01:04:26:20 - 01:04:48:23
Unknown
And there was an HBR article that you, I can't remember if you put this in. No, actually, I found this on a Harvard, business review that has a little chart here, where they looked at the top ten. Jenny. Jen, I use cases in 2024 compared to 2025, 2024. The top use case was generating ideas that dropped to number 6 in 2025.

01:04:48:23 - 01:05:19:11
Unknown
The top of what they're saying is therapy and companionship. And I think that's that's an interesting, kind of place that we find ourselves with these chat bots as they continue to proliferate. Yeah. We had two headlines in here. One was from Wall Street Journal with AI chat bots linked to psychosis, say, doctors. Yeah. A little, which I think, but basically what the story says is people already had these problems, and then they were using the AI, and the AI validated what they were saying.

01:05:19:13 - 01:05:39:09
Unknown
At the same time, the Guardian, which tends to be and I love the Guardian, but they tend to be, puritanical about technology these days. I had a story saying, could AI relationships actually be good for us? So it's like, is red wine going to kill you or save you? And it's totally last year. That's totally it. Yes.

01:05:39:12 - 01:06:03:12
Unknown
There's there's so many more. I'm sure you all are screaming a million things at us, but, what about a quick how about next? Yeah, yeah. Okay. I'm just gonna throw these out here real quick because I know we're we're going a little long at this point. I think I think this next year is there's going to be a little bit of a reckoning because I'm kind of seeing it a little myself as well, reckoning around how productive we actually are when we use these tools.

01:06:03:18 - 01:06:24:08
Unknown
Like the more, the more our eyes open, as at least I can speak for myself, the more my eyes open to the possibilities, the more I'm like, oh well, I want to do more using these tools because that's going to enable me to do so much more. But then I realize on the other end that it just it creates a different type of work that I have to then do to make sure that the time that I saved over here is okay.

01:06:24:08 - 01:06:40:13
Unknown
And I kind of think in some ways it's kind of a neat equal to a certain degree. And I wonder how how more people are going to come to terms with that. You know, this was magical at first. Now I'm starting to really get a clear picture. Is it actually making more productive, or is it just making my my work different?

01:06:40:18 - 01:06:57:15
Unknown
And I think people might see that a little bit more, I shopping moving beyond the demo again that that has a big trust component. But I think we're early on the AI shopping thing. And if we know anything about the internet and how it all works, that's that's pretty critical that that moves the industry in a lot of ways.

01:06:57:15 - 01:07:14:21
Unknown
When they figure out these new ways to buy things and to sell things, and AI is squarely in the in the crosshairs there. I remember I put it, I worked for company ask way back in the day, there was the belief that people were not going to use the levers. People are going to use the internet to buy things.

01:07:14:21 - 01:07:35:28
Unknown
You're not going to trust them. Obviously, I remember that. Then they weren't going to buy expensive things. And we started a style.com store with Neiman Marcus with very expensive things. And so these things change over time. I think you're right. The shopping stuff will get real. I just have you use Rufus at Amazon. I have that's where it, like, suggests certain things comes up in the sides.

01:07:35:28 - 01:07:55:24
Unknown
Right. And I've been ignoring it. Yeah. This is more than you don't want to know about me. I have seen it. It's cold here, so I needed a new robe that was warmer. And I went to the store, and they were like, you creepy feeling. And I, and I, and I looked at Amazon, look at robes and, so Rufus came up and I said, I don't want, a robe that feels creepy and too soft and all this.

01:07:55:24 - 01:08:15:19
Unknown
I want something that feels like a towel said, oh, you don't want this, you want that? And it made recommendations. And I thought, oh, okay. I got, I got something in the and the shopping cart as we speak. Okay. Well, I'll be curious to know if this feels like a towel or something. Creepy. Yeah. Okay. Creepy robe.

01:08:15:22 - 01:08:36:24
Unknown
And, I think my my last one is just in, like, I'm curious about this kind of growing. Let me see. How do I phrase this? I think we're going to see a lot more AI hardware, AI driven hardware, right. And a lot of it is going to be built around this idea that as as we as users work with AI, we realize the value of context.

01:08:36:24 - 01:09:05:08
Unknown
I know I do, and so I realize the value of having certain conversations and, and, meetings and whatever recorded. So that I can then work with that data on a comprehensive way after the fact. But what that opens is this kind of like, growing kind of, connection into, in a more openly surveillance world where people want to be recording more things, their devices enable them to record more things because they're wearing glasses that make it really easier.

01:09:05:08 - 01:09:26:16
Unknown
They're wearing a ring that makes it really easy. And how how does the how do other people in the room feel about that weighed with the kind of the, the, the reality that having these recordings can be really helpful and useful in how we do things. And so, I don't know, it's it's a weird kind of conundrum state.

01:09:26:18 - 01:09:45:16
Unknown
So by for next year real quick, the first one, I'm going to agree with your first one, which is that I think that are waiting to see us whether we're productive. And I think at the larger level, in an industry level, is there real utility and real revenue at a scalable way? Is this really a business or not?

01:09:45:16 - 01:10:05:12
Unknown
Right now it's so hot and people are buying chips and they're building centers. But is this really going to take off or not? And I think it probably will, but we'll see whether the podcast sticks around as a result. To number two is kind of coming back to the China theme, just generally, will OpenAI stumble or will it, recover and and be a real competitor right now?

01:10:05:12 - 01:10:20:23
Unknown
I think they're stumbling. And then number three, going back to our earlier conversation in the show, I've been thinking about this this last week or so, and I was thinking that that AI does have cooties. Witness the stories we talked about, and I wondered whether people were going to come to a point where they would rebrand AI.

01:10:20:29 - 01:10:43:10
Unknown
They would stop calling it AI, they start calling other things. So so happens I this week I listen to a new Books Network podcast. It's a podcast of all kinds of academic books, really long, interesting, but pretty in-depth things. So it was a discussion with Thomas Hague, who's a University of Wisconsin Milwaukee professor, who is a professor, historian of computers of technology.

01:10:43:10 - 01:11:07:25
Unknown
He wrote a book about the history of computers. He's writing one about the history of AI as a brand. And in the podcast, interestingly, he stole my thunder because he's smarter than I am. And knows there. And he said he predicted that probably this year, coming up a year or so, people will start calling that machine learning again or, or other things rather than I if I takes on enough of a negative connotation.

01:11:07:28 - 01:11:28:00
Unknown
So that'll be interesting to see it. Yep, absolutely. Although machine learning is a little bit more of a mouthful, it is two letters, but yes. No, I think you're absolutely right though. I think there is. There is a little bit of that, that Cudi factor that that might make that need to happen or make people feel like it needs to happen.

01:11:28:00 - 01:11:47:14
Unknown
Anyways. I'm sure we've missed a lot, but that's that's a couple of things that we're looking towards in the year 2026, outside of the fact that we're just going to continue doing the show and, continue growing it. And I can't think of a person who I'd rather do this with on a weekly basis. Jeff, I really appreciate doing this show with you.

01:11:47:16 - 01:12:19:01
Unknown
I learned a ton. I learned from you, and, I just appreciate your time. Thank you. Thank you. Jeff jarvis.com is the place where you can go to find all of Jeff's works, his books primarily hot tape. Coming soon, available for preorder right now. Right on Comingsoon Gutenberg parenthesis now magazine now. And the web. We we've right now Jeff jarvis.com and then I inside dot show for all of our episodes including our previous episode where we interviewed Ben Gerstle.

01:12:19:01 - 01:12:40:13
Unknown
Again don't miss that if you did already. It's there for you I inside dot show and then finally Patreon.com slash AI inside. You can go there. You can, follow us and support us on a deeper level. And we appreciate if you do that. And, we also have some amazing, executive producers who support us on a deeper level each and every week.

01:12:40:16 - 01:13:00:16
Unknown
Doctor du, Jeffrey Marie, radio Asheville, 103.7 Dante. Saint James, Bono, Derek, Jason, Nyfair, Jason Brady, Anthony Downs, Mark Archer and Carsten. Smash. So thank you, everybody. And I'm covering up Jeff I apologize about that. Thank you everybody for your support of this show. We could not do it without you. I know we say that a lot, but it's absolutely true.

01:13:00:19 - 01:13:16:19
Unknown
And then next week, just real quick before we end things, I'm going to be in Las Vegas for the concert. We didn't talk about show. We didn't talk about that. Mike. Yeah, so we'll talk about that off air, I'm guessing probably to next week's episode is going to be a little different. Like I might while I'm at the show.

01:13:16:19 - 01:13:31:29
Unknown
Find a couple of people and do some interviews and everything. Just because my schedule is going to be super packed, you're going to get an episode. Don't know what the format of that is going to be yet, but rest assured you'll get an episode and, that's all there is to it. So thank you everybody. Anyway, germs there.

01:13:31:29 - 01:13:47:26
Unknown
Stay away from the white man. Yeah, I'll have my, my my hand solution. Hopefully it'll do its job. Thank you, Jeff, for a wonderful year and looking forward to next year with it back. And thank you everybody for watching and listening. We'll see you next time on Ironside. Happy New Year.