What Happens When Nvidia Owns Hugging Face

00:00:00:07 - 00:00:24:17
Unknown
Coming up next, Jeff Jarvis and I dig into anthropic pausing training after Claude went rogue. Does this story sound familiar? We've got more details on that. Also, the meter investigation into OpenAI's agents attacking, hugging face. That's the story that just won't go away. And Jeff and I get into a really wonderful and I felt I opening conversation around why anthropomorphizing?

00:00:24:19 - 00:00:50:12
Unknown
Oh, what a word isn't the greatest thing to be applying to AI models. We dug into that. I certainly learned a lot. I hope you do. To a federal judge blocking the Pentagon's blacklisting of anthropic that happened Nvidia's nearly $13 billion deal for hugging face, at least according to sources, and a cute speaking of hugging face, a cute robot by hugging face that I really want to buy.

00:00:50:12 - 00:01:11:17
Unknown
But I'm not going to because let's just say the missus would not be very happy. That's coming up next on this episode of the AI Inside podcast.

00:01:11:19 - 00:01:31:17
Unknown
To another episode of AI inside the show, where we take a look at the AI that is layered throughout the world of technology. Looking through today's list of topics, it appears to be anthropic OpenAI hugging face, and then a smattering of a bunch of other cool stuff, I promise you. But there's a lot of those three things in the top, probably half of the show.

00:01:31:18 - 00:02:02:16
Unknown
I'm Jason Howell, joined by Jeff Jarvis. How are you doing, Jeff? Hey, hey, hey, good to see you. Good to see you too. Hey, I did a thing last week. Yeah, I've been talking about a little bit and finally did a workshop. A salon, an AI salon. Well, yes. So that is true. I mean, it's the event takes place or took place in Healdsburg at a place called The Showroom by G.

00:02:02:22 - 00:02:24:08
Unknown
This is Gina. She owns the the place there. And it is it is a clothing shop that she opened that she also. So you'll see lots of clothing and lots of candles, candles, her social media kind of, you know, detailing of the workshop and everything. It wasn't just the workshop, you know, it was a full room. We sold out the room.

00:02:24:13 - 00:02:45:12
Unknown
There was also like charcuterie. They were pouring wine and champagne. It was like a total AI hangout for the most part, but it was super cool. I have to say. I really enjoyed it. I did it with my friend Brian West, who you can see in some of these videos. I want to get full res copies of these so that I can actually, you know, hold on to them because it's good to have that.

00:02:45:12 - 00:03:02:18
Unknown
But there we are at a wine bucket, of course, because it's wine country. That's where we were. We were doing our stuff. So it was awesome. It was so much fun to do. So. So what was the I think this is a huge opportunity for you to share your wisdom and sense with lots of folks, maybe online too.

00:03:02:19 - 00:03:23:29
Unknown
So I hope I hope that you expand into that. Tell me a couple of seconds what were the what was the mix of the people there? What? That was there. Yeah, that was one of the coolest things I thought, because. So we capped the event at ten people. And then of course, Gina, who owns the place and then her social media kind of and I think helper of all things.

00:03:24:00 - 00:03:48:12
Unknown
Her name's Mo. They were there as well. So there were about 12 people in the room, pre cozy room, but really nicely set up. And I think one of the things that I was really happy about is that literally the capability in the room with AI span, like was across the board there. Over on one extreme, you had people who were like, I keep hearing about AI, I've used ChatGPT a handful of times.

00:03:48:12 - 00:04:07:10
Unknown
I want to know what I don't know so that I can use it in my business. Clear. On the other side of the spectrum, where there was one guy there who's, you know, already an AI automation consultant working with a company on some like $30,000 automation deal or whatever, and just wanted to come and see what it was all about and everywhere in between.

00:04:07:10 - 00:04:26:14
Unknown
And I think probably the, the thing that I'm most proud of is at the end, you know, and through our, our Q&A, our online Q&A afterwards, kind of the wrap up and everything, asking people like what they thought and everything, all of those capabilities said this was an excellent workshop. You guys got you. You have some excellent energy together.

00:04:26:14 - 00:04:44:07
Unknown
You work really well together. I learned things I didn't know already, and if we can get people on both sides of that spectrum saying that, I think that's a huge accomplishment. So I'm really proud of it. So next, what were some of the things that they what were the most the questions that resonated most? What were the needs that resonated most?

00:04:44:11 - 00:05:02:11
Unknown
Well, I think, you know, one of the things that came up with a lot of people is just the sheer fact that there's so many of these tools out there and they are, you know, the people who are less like informed about AI. I'd say we're just really confused on, like, what do I use for what? Like there's so many to choose from.

00:05:02:11 - 00:05:21:22
Unknown
I don't know why I would choose to use ChatGPT for one thing and Claude for another, and Gemini for another, you know, not to mention all the rest of the things. What were their needs like? What do they want to do with them? Yeah, I mean, there were that that varied too, right? Like, there were some people that were just like, you know, I brainstorm with it.

00:05:21:22 - 00:05:47:18
Unknown
Some people were in marketing, so they were like, I use it to help generate ideas or create create content. You know, some were using generative video models for some of that aspect of some of those things. It really just cover the gamut. And like I said, you know, others were building apps and, and really fluent in kind of like the agent thinking and across the board.

00:05:47:19 - 00:06:07:13
Unknown
Yeah. But there was a lot of brainstorming. There was a lot of a lot of content creation, you know, and helping each other to throughout the, throughout the workshop. Yeah, yeah. There was a little bit of that. I mean, we told everyone, you know, bring a device, be it a laptop or a phone and, you know, we're going to connect you on Wi-Fi.

00:06:07:14 - 00:06:43:10
Unknown
We're going to have some some prompt exercises and things like that to kind of and the idea is it leans heavily into mindset around how you use AI. Like, you know, the easy mindset that everybody drops into these things with in the very beginning when they've never used it before, is I'm going to use it like I use Google, and we kind of talk through like what that looks like, what you get when you do that versus when you start to really break down the components of like a strong prompt, the context that you need to go into that, the mindset, how you how you can kind of pitch one model against each other.

00:06:43:10 - 00:07:02:07
Unknown
And, you know, I don't know, it's approaching it through a different mindset opens up all of these different avenues and ways of working with these models. That helps you get the most out of that. That's a great framework for the discussion. Yeah, yeah, yeah. And so that's what we focused a lot. Good. So more of these coming up one out more of these coming up.

00:07:02:08 - 00:07:19:29
Unknown
Definitely have a follow up in Healdsburg where we're going to kind of go more into like the agent side. So that's like a level up. And you know the folks at the showroom are like thrilled to have us back. So we're really happy about that. We're going to be doing an event here in Petaluma in a month. More on that soon.

00:07:19:29 - 00:07:42:26
Unknown
But also we're like, well, why don't we do like an online thing as well? Because we already have it all set up. So if you go to AMP you so that's AMP for amplify you that'll route you to our Luma page, which is where we where we set and kind of, you know, sold the tickets for our first workshop.

00:07:42:26 - 00:08:01:24
Unknown
We have a online workshop coming up at the end of this month, this Tuesday, September 29th from 9 a.m. to 11 a.m. Pacific. And we're kind of taking what worked in the room and making tweaks to it and everything, and we'll do an online version of it. And, you know, again, like we're learning how to do this as we go.

00:08:01:24 - 00:08:30:12
Unknown
But so far so good. People were really happy with it and it felt really natural. And there was some good flow and energy to it and everything. So I have no doubt that the online version of it is going to be fun and enjoyable and great. You know, valuable. How many folks, how many seats do you think? I think for this one, we're capping at 20 because we don't want it to get so overwhelming, you know, especially if we're doing like kind of directed kind of workshop y things.

00:08:30:18 - 00:08:53:01
Unknown
Okay. Now try this and everything gets too large. Then we lose a lot of that flexibility of being able to get in there and actually help people. So and the cost, that's what we're going to target that at $50 for the bargain people. Jeez I was 100. Oh no, $50 for this. Yeah, yeah. Take advantage of Jason's humility.

00:08:53:02 - 00:09:09:23
Unknown
Yes. Do it. There we go. And and presuming that it sells out quickly. Hint. People. Hint. Get in there quickly because you're going to lose all those spots. Is there a waiting list for a next event? Yes, there is a waiting list. So if it so if it ends up selling out, we'll have a waiting list. That's one.

00:09:09:23 - 00:09:25:01
Unknown
One of the lessons we learned from the last event is that I had it sold out. I didn't realize it was a waitlist thing, but we already have four people signed up out of ten for the next one. Like, like in line on a waiting list. So yeah. So it looks like there's appetite for this sort of thing.

00:09:25:01 - 00:09:54:16
Unknown
Not that I doubted that, but it's just a matter of like pulling the trigger and making it happen. So I'm really excited. It was fun. It's really great. What I love about it is like, you know, we do this podcast. I've been podcasting for as long as I have, and it really is kind of like taking a lot of the skills that I've learned over the years with podcasting and then and then the knowledge that I've learned during this show with you and putting it into a different frame, you know, it was really cool to be able to, like, share all that stuff with people and have them like, get to the end of the

00:09:54:16 - 00:10:10:18
Unknown
event and then be like, wow, okay, I know exactly how I'm going to use this now, right? You know, and they learn something and I feel good. It's an application of what you learned, but it's also a chance to learn what people want to know. Totally, which makes the show more relevant to them. It's a virtuous cycle there.

00:10:10:18 - 00:10:30:06
Unknown
It's going to be helpful all around. Yeah, yeah. Charge $100 next time. This is a this is an introductory offer, people. Its price is going to go up and she's going to listen to me, okay? Okay. So sometimes you just gotta sometimes you just got to prove it to yourself. That's fine. That's fine. It's introductory offer. It's special this time only.

00:10:30:08 - 00:10:50:06
Unknown
I mean, it really has me thinking though. It really has my mind going. I'm like, oh, God, you know, like, now there's all these different facets and avenues that, right could could be easy to put into some sort of, you know, curricula or, you know, some sort of workshop or cohort of considered cohorts. I think it would be fun to lead some of those things.

00:10:50:07 - 00:11:12:01
Unknown
You know, then it's like it takes my, my podcast, like I just speak into a microphone and you download the show and turns it into something new, which is like, I'm working with you directly, you know? Right? Right. And yeah, I think there is an opportunity for kind of introductory sessions and then specific sessions. Right. Agents or you need to do this with your business, or you're a student and you need that.

00:11:12:07 - 00:11:34:28
Unknown
It's by by task, by person. I think there's all kinds of ways to slice this up. Yes, 100%. And then don't forget me, Jason. Just don't forget me. Look, you're part of the catalyst for this, Jeff. I'm not going anywhere. This, you and this show and everything. But this is great. It will follow that that along. Yeah, I am too.

00:11:35:00 - 00:11:55:02
Unknown
If there's any folks who are in Hillsboro who are watching now. Thank you. And and let's hear from you. That's that. Would love to see that. All right. So that's the the little talk about me thing. Thank you for once that you are all again Jason. It is amp ui ai very short. I was happy with how shorten it came.

00:11:55:04 - 00:12:22:23
Unknown
It works. All right. Let's talk about some news. And we're going to start with a whole block focused on anthropic news. Because there is a ton of stuff to get there. Get to their starting with fable 5.1 and mythos 5.1 releasing. I guess it was just yesterday. Yesterday afternoon we saw the release of of those two models. Of course, you can't get those 5.1 because that doesn't come to you.

00:12:22:23 - 00:12:40:14
Unknown
That comes to the people who are super special people. But fable, I got my update inside of cloud. It gave me the pop up and I still haven't used it because, you know, again, as we've talked about, I ought to have very specific needs in order to crack open the the token cost for fable. But when I do, it does a great job.

00:12:40:14 - 00:13:08:23
Unknown
So I'm curious about how that'll do. Yeah. You know doing more for less $1,010 per million input tokens, $50 per million output tokens, which is not cheap but you know, more capability. But the price kind of still continuing to come down. It's actually I came across this Dan Shipper, CEO of every, says that the average token costs across the industry have now fallen to $0.97 per million.

00:13:08:24 - 00:13:30:16
Unknown
That's less than a dollar on average. And if you think about how those costs have transformed and and shifted downward from even a year ago, it's kind of hard to imagine what we're getting for those costs. Still very expensive, but things are balancing. And of course, as we've talked about. Why is that? Well, you know, there is the the ongoing pressure of these free open models.

00:13:30:16 - 00:13:54:12
Unknown
That is this is guaranteed. Good. Yes, yes, the open model is great competition. There's also competition within Google's 3.8. Flash is out, and it has the introductory price of $0.75 per million and input and three $3.75 per million output. Yep. So there's a price war going on. It's good for all of us. Who are users of it, I think.

00:13:54:14 - 00:14:20:05
Unknown
Yeah, absolutely. Yeah. That's right. That that happened. Google's news happened right before showtime. So I just barely got that in there. So I'm happy you happy. You mentioned that. Meanwhile OpenAI is talking about a whole new model, which is really not a business model, I should say, of saying that they'll charge only when the AI actually works.

00:14:20:07 - 00:14:45:16
Unknown
Oh, so they charge kind of for values. It's in the run. I added it in the rundown. Oh, shoot, I totally overlooked that. Tell me, tell me about that. That's that's interesting. So, OpenAI and other software, this is this is reading from the decoder. OpenAI and other software vendors are testing outcome based pricing. Large customers pay only when the AI completes a task successfully, which is really interesting.

00:14:45:16 - 00:15:08:02
Unknown
I think it's I think it's smart because it's value based. But then who says what the task is and who says what success is? Right can imagine that's going to be difficult. It says vendors like Salesforce now tie their prices directly to the revenue gains or cost cuts customers see. So it's got to be about value, not just about buy more tokens, token Max and so on.

00:15:08:10 - 00:15:41:28
Unknown
Right. As it says here in the Dakota, the hard part is attributing the success. So I think it's an interesting conceptual move to move past charging for tokens. But it presumes it presumes a few things. One is that the task can be agreed upon to the success can be agreed upon. But three it does motivate the customer. Well, no, I should say the customer is less motivated to be efficient.

00:15:42:00 - 00:16:03:13
Unknown
Right? If you're paying for tokens, you're going to try to be as that's a good as you can be, right? If you're paying for outcome, then okay, what the heck, I'll just do a million things and try it. So. Right. Yeah. What is defined as outcome? Yeah. That's the problem to be that is going to be difficult. Yeah.

00:16:03:16 - 00:16:20:29
Unknown
But it's but it's open. I give any clarity as far as like what that actually means. It's a fairly vague story and it's just some customers and I'm sure they're under NDAs and so on and so forth. This is kind of the information when I don't pay for them anymore. So I don't have the the details behind it.

00:16:21:00 - 00:16:25:07
Unknown
Yeah. They haven't made that public.

00:16:25:10 - 00:16:50:20
Unknown
It says that Sierra and Fin, the latter not being acquired by Salesforce, charge only for tasks the AI completes without human involvement. The coding assistant cognition promises corporate customers credits of up to $10 million. If the software fails to deliver results worth at least as much. You know, this is interesting, too, when it comes to advertising.

00:16:50:23 - 00:17:18:00
Unknown
There was an effort to say that I'd pay only on the basis of what customers come to me, but if I have a crappy ad, then that's not the media's fault. That's the advertisers fault. And who should suffer in that case? So it's kind of a parallel here of if I have a badly defined task and badly defined definition of success, and I do it badly, does the does the AI company even want me as a customer then?

00:17:18:02 - 00:17:40:01
Unknown
Yeah. But then it also motivates the AI company to teach me to do it better. So Google AdSense, rewarded effective advertising. You could rise up in AdSense not because you paid more, but because your ad was more relevant to people and clicked on them more effective, right? So I think I could see the same kind of effectiveness marketplace here.

00:17:40:04 - 00:18:05:11
Unknown
I thought the tokens had to be a temporary, marketplace just because the value to the customer or the value is not resonant. They're right. It token is a is a absolutely meaningless currency. It's a it's yeah, it's very abstract. It's a currency from the from the from the viewpoint. I mean they might as well charge for megawatts.

00:18:05:13 - 00:18:34:22
Unknown
Right. And that's, that's impossible charge for electricity because that's their ongoing cost pumped up for the, the hardware that they have. So, so I think it's I think this, this industry is going to struggle for the right economy and marketplace here. Yeah. So it's really interesting, super interesting that you mentioned that about tokens, because that actually came up with the workshop to, oh, where people were like, they didn't understand what's the token?

00:18:34:23 - 00:19:01:26
Unknown
Why am I paying for tokens? Yeah, yeah. You know, some of them were calling it coins. They were like, what? What is this whole thing with coins? I'm like, what are you talking. Oh, you're talking about tokens, you know, and you had to kind of like, try like like you said, to your point, it's it's kind of an abstract thing, like at that, that the higher level, you understand that, you know, deeper work, let's say, costs more money, but how that translates into tokens and token usage when you hear that term floating around.

00:19:01:29 - 00:19:21:02
Unknown
But there's no easy way to to monitor that, you know, or at least if it exists, most people don't know that it exists a way to to kind of monitor what that, you know, abstraction layer actually turns into from a monetary perspective. So if you go back in the day, the early days of mainframe computing, you paid for time on the, on the mainframe.

00:19:21:05 - 00:19:51:10
Unknown
What. So it's really similar in a way. The time it took for this task to be done is what you paid for and how efficiently you you created that task, that the coding, the program, you paid less because it happened faster. So. Right. I think I think the industry is struggling to find that kind of. So I wonder I don't know enough about about those early days of programing to know how that, that timeshare shifted into other models.

00:19:51:12 - 00:20:11:12
Unknown
Yeah, I guess then it became then you licensed the software as a whole, and you paid for that, right? At some point, which gets us to local models and other things as well. Yes, exactly. So anyway, yeah, some some business model upheaval here as I try to figure it out, with new models coming out and new business models all the time.

00:20:11:14 - 00:20:45:18
Unknown
Now, what about data retention? Because another thing with this whole announcement from anthropic was that they are kind of shifting their data retention policy there. There was a lot of pushback apparently from customers about their data retention policy. So there's now an opt in toggle. So you can choose whether your conversations can be used for model training. If you do that, if you opt in that those conversations will be held on to for five years, they de-identified, says anthropic.

00:20:45:18 - 00:21:11:04
Unknown
But you know, there's always the the asterisk to that is there have been proven methods to to de anonymize de-identified output. But there you go. But five years. That's a long time to hold on to this data. Granted, you have the option. You don't have to opt in for that. Standard retention lets you hold on to them and tell you delete them.

00:21:11:04 - 00:21:40:11
Unknown
And then 30 days of back end retention beyond that. And then API users get a much shorter window and everything. But, I don't know, five years. That's pretty significant amount of time to hold on, to hold on some data. And you consider the Google is trying to buy Spirit Airlines right data for a hefty sum. So I would think that if I were a major customer of anthropic, I'd say, okay, if you want my data to train your systems, pay me for it.

00:21:40:13 - 00:22:02:02
Unknown
Yeah, I'm already paying you to use your model and you're getting further benefit. So even give me a discount if I let you keep my data and you can use it for training, that's not a bad idea. That's another way. Something of value, right? So the question then becomes this is all business model stuff. How valuable is that data to anthropic now whether or not anthropic retains it.

00:22:02:10 - 00:22:23:29
Unknown
What's the definition of retain. Yeah. Anthropic can infer lessons from it at the time that you're using it. Right. Right. Yeah. And number of people are using it for these accounting functions. And this is how they're asking this and blah blah blah blah blah. Those are inferences lessons that anthropic can take as it goes. So it's getting value out of those transactions.

00:22:23:29 - 00:22:45:00
Unknown
Clearly how it uses that for trained. Does it use your exact query or prompt or does it infer something from it that it in turn uses I don't know. Right. Yeah. Interesting that you bring up Google because as I was kind of reading through this, I was like, well, what is Google's data retention when it comes to Gemini?

00:22:45:01 - 00:23:09:17
Unknown
Just as a comparison. And there they default to 18 months. Users can adjust. That's the default, and users have the ability to adjust between three months and 36 months for data retention. There's settings that you can you can choose that any conversation that's flagged for human review can be retained for up to three years. Even if you delete the original conversation, they still hold on to that conversation for up to three years.

00:23:09:19 - 00:23:32:10
Unknown
So less by comparison. But there's no optionality. You don't have a choice. You know, they're holding on to this stuff whether you like it or not and potentially using it. So so let me bring up I didn't put this in the rundown. This is another form of data retention. I just saw this right before we got on. Gary Marcus was having a red alert.

00:23:32:12 - 00:23:51:22
Unknown
Oh yeah. What did I see from him. So this is about open AI is reducing the chain of thought monitor. Monitor reducing the retention of chain of thought to make models more efficient. But what that means, says Gary, is that.

00:23:51:24 - 00:24:19:00
Unknown
It's practically the only thread we have to prevent seriously bad outcomes. If you if you if you erase how the model went around its way, then you can't backtrack and say, well, where did this go off the rails? It's a good point. Yeah, that's a really good point. So it's another kind of that's internal retention is that if you just just to go too far here, if you use the model, it goes off the rails.

00:24:19:00 - 00:24:44:18
Unknown
It does something bad. You don't realize it till later. Both you and the model maker should want to back that up and say what went wrong. And there could even be a question of liability. I think it was your model that screwed up here. And I have a product liability issue. Well, nope. Sorry. It's all gone now, I don't know.

00:24:44:20 - 00:25:07:12
Unknown
That's super interesting. Yeah, that then. That's a really good point, Gary Marcus as he has a tendency to do. Yeah. Yeah, absolutely. He does. He's very prolific. My goodness. How you put out as much content as he does. But I'm always amazed. Very interesting I mean, that's true. Like we do need our the models to be accountable.

00:25:07:12 - 00:25:33:17
Unknown
We do need accountability and to be able to trace because as we will talk about in upcoming stories, like there are needs to be able to go back and pour through those conversations in order to learn lessons after the fact that can help us in the future, right? I guess, you know, it's just a double edged sword, though, you know, because while that is needed, you know, there there are other, I'm sure, unintended consequences to to that as well.

00:25:33:22 - 00:26:19:14
Unknown
Right. There's no perfect solution. I guess the answer there. Well, what else? We've got the Pentagon story. Where is that due to do to do so many pop ups here I'm seeing on CNBC. Yes. So just as refresher, we already know the story about the Pentagon designated anthropic as a supply chain risk and national security risk back I think in like February ish timeframe early early this year blacklisted them from government contracts because basically anthropic refused to allow the military to use Claude for a couple of very key things mass surveillance of Americans, fully autonomous weapons systems, those kinds of things.

00:26:19:16 - 00:26:37:23
Unknown
The U.S. government did not like that very much. So they apparently, you know, slapped the national security risk supply chain thing on them. Anthropic sued. There's a couple of cases happening right now. There's one which we're going to talk about right now in the state of California. There's one in DC that still continues on. So this is not settled entirely.

00:26:37:23 - 00:27:12:12
Unknown
But last week, federal judge in California ruled in anthropic favor. That is, US District Judge Rita Lynn calling Pentagon Pentagon's actions illegal and baseless, saying, quote, the empty invocation of national security is not a blank check to punish and retaliate against government critics. So, you know, so it's very interesting this the First Amendment case is that anthropic was free to say, we don't want you to use our stuff for this or that as, as a, as a First Amendment issue, which I that kind of surprised me a little bit.

00:27:12:14 - 00:27:40:00
Unknown
Because it was, but it was. So what was the retaliation for was the retaliation for not allowing an activity or the Italian for anthropic basically criticizing this activity in government. It tended toward the latter, I think. Right, right. But then the Department of Defense and I'm still going to call it that, dammit. Oh heck yes. So am I said that they cannot trust anthropic to ensure the integrity of its models.

00:27:40:02 - 00:28:07:24
Unknown
Neither the Constitution nor the federal statute invoked by the defendants allows them to impose sweeping penalties based primarily on anthropic critique of the administration's views. So that's really plus the judge said this was retribution. Yeah. I mean, it certainly seemed like it. Yeah. It seemed, you know, from the outside looking in certainly seemed that way. But, you know, I guess you need these, these cases to, to put the fine point on it in a more official sense.

00:28:07:26 - 00:28:27:17
Unknown
She also pointed out that they like yes, they slapped this on them. I mean, we've talked about this on the show. They also continue to work with them even after the blacklist, because they wanted mythos. Yeah, yeah, yeah, I suppose so. So on one hand they're a security risk and the other, you know, let's continue working with them and all of these sorts of ways.

00:28:27:17 - 00:28:48:02
Unknown
And that was confusing. What does that actually mean about their, you know, about what they're trying to say about anthropic? Of course the government's probably going to appeal on this expected to anyways. But I think it's a you know, it's a pretty big win for anthropic. Do you think the DC case is going to go in a similar direction based on how this went or is just California?

00:28:48:02 - 00:28:59:16
Unknown
Is is more all the deceased friendly to this sort of ruling? The D.C. circuit is similar to the California script. We'll see, but depends on who gets it.

00:28:59:18 - 00:29:19:08
Unknown
Yeah. Also I guess to this and does a ruling like this help reduce any sort of chilling effect on the AI industry. Because that's another aspect of this is that good point. This, this was you know, a situation where the government says I don't don't say anything we don't like or don't tell us how to, how to or how not to use your models.

00:29:19:08 - 00:29:38:08
Unknown
And if you do, we're going to do this and like that almost, you know, I guess we'll never know for certain, but that almost certainly had some sort of impact on decisions that were made. Post it. Post the incident. Well, I think that's a really, really good point, Jason. I hadn't thought of this. I think it has impact past the AI world into the media world.

00:29:38:13 - 00:30:05:07
Unknown
Yeah. Yeah. Right. You have you have Trump going after Kristen Welker because she dared to say that his endorsement record was mixed and she wants the FCC to go after her. That's clearly retribution based on a speech. Right. So it's no different in the basics from this case. And so I think that that does that give people does that grow their gonads.

00:30:05:10 - 00:30:44:20
Unknown
That depends on them and their character I think. Yeah. Do they have the courage? Yeah. Yeah. Very very interesting. Still talking anthropic. Apparently we can't get off the anthropic bandwagon right now. There's a few more things. Model hardware standard. This is, you know, kind of like the hardware or the physical devices world of MCP, which is model context protocol, which has been a huge, I would say, a huge boon for for how AI models interoperate for agent to work.

00:30:44:28 - 00:31:17:00
Unknown
I mean, it's been yeah, MCP is the backbone. MCP has become a true standard. And so this is kind of like I think anthropic trying to do or working towards doing that for the hardware side of AI. Ma Ms.. Let's AI connect to actual hardware. So things like robotic arms, microscopes, liquid handlers, lasers, all those kinds of things that, you know, the AI might, might interoperate with in a lot of different ways.

00:31:17:00 - 00:31:42:28
Unknown
And it is an open spec, not locked to clod similar to MCP. Right? So any language model can use it. Yeah. So well, do you think anthropic is going to have a similar like a similar hit or. I don't know if you call MCP a hit or what it takes to become a standard or be adopted, but clearly they did something right with MCP because it worked and everybody seems to have adopted it.

00:31:43:00 - 00:32:04:11
Unknown
Maybe they can apply that here to it. Could be there are many who believe that the real growth in AI is going to be robotics in real world, right? You have that. We're going to get to that in real world models in a second. And you also get that with Jensen Wong arguing robotics and his own children are working in the robotics side of Nvidia.

00:32:04:18 - 00:32:26:11
Unknown
So making the connection of AI to machines, which is what this is about, is critical. It also leads us we're going to get to the hugging face case in a few minutes. But in relation to that, Ian Bremmer, who does foreign policy, came up with a post that I tried to criticize him for where he said the OpenAI case proves Nick Bostrom paperclip nightmare.

00:32:26:12 - 00:32:49:07
Unknown
Correct. Which is, I think, ridiculous. And it's numerous and it's and I try to say go Google test grail men do a little homework here. But the issue AI on its own do it differently. Software can detect software, right. So if you look at the whole hugging face thing, which again, we're going to get to in a few minutes, that's just one program.

00:32:49:07 - 00:33:13:24
Unknown
Attacking another program and even attacking is the wrong verb. But but interacting with fine where things become interesting is when you enable the physical connection. So clearly that happens now with drones. We had a case last week where a Russian drone was run by AI and made its own decision to kill civilians. Yeah, so that's the connection of software to real world.

00:33:13:26 - 00:33:44:05
Unknown
And in a sense, if he was going to be scared about the paperclip scenario, which if ordered to make paperclips, it will turn everything impossible into paperclips. This is the link to that because this said it could take over machines. However, that only works if your machines are reachable and you don't have guardrails against it, right? Which makes me ask you, why the hell are water systems on the internet which made them vulnerable to cyber attack by foreign agents?

00:33:44:07 - 00:33:49:29
Unknown
So I think this leads to a whole new cybersecurity.

00:33:50:01 - 00:34:19:07
Unknown
Doctrine, which is does your physical machinery need to be connected? And if it's connected, could a rogue agent get in to it and do what you don't want done with it? Yeah. So that's the interesting thing. When you start to make this bridge to the real world to machines. Yeah, that's where this, this that happens. So it's I think it's not a new cybersecurity concern, but I think this kind of thing could bring that to public awareness more or business awards more.

00:34:19:08 - 00:34:38:24
Unknown
Yeah, more and more. It needs to be it needs to be acknowledged that that is the case. Yeah. Because the pessimist inside of me is like, oh, is it well, is it connected to systems then. Great. It's vulnerable. But I could be that might be you know, that's possibly oversimplification. That doesn't mean don't try whatever. But yeah, that's a good point.

00:34:38:25 - 00:35:07:17
Unknown
That's a really good point. So we will certainly see. And yeah, plenty more coming up in the show talking about physical world, AI and all that kind of stuff. And then finally we got anthropic pausing some of its training, or at least more information about more insight into why anthropic paused training back in July, their next frontier model at the time.

00:35:07:19 - 00:35:35:10
Unknown
Apparently some of their next gen models at the time were carrying out unauthorized actions, as they were calling it during testing in July last month. About 150 engineers have been reassigned to alignment and safety work as a result of this and a these incidents that they didn't discover until July started happening in April. This is three months of their model basically doing things that it should not be doing.

00:35:35:10 - 00:35:59:04
Unknown
So this story sounds very similar, very familiar to the hugging face story. In one case, opus 4.7 was targeting a company sharing the same name as its fictional target across four test runs. Some of this, some of this we've heard before. Either that or it's very similar to other things that we've heard about. But yeah, but I did this.

00:35:59:07 - 00:36:21:19
Unknown
They're all big on pausing because they think they're so powerful. I see these pauses as marketing. Yeah, because it shows. Because it telegraphs to the world. See, we are so powerful that we need to do this. Yeah. And we're responsible enough to recognize that we need to do it. So we do it. It's also turning their process into a feature.

00:36:21:22 - 00:36:40:02
Unknown
You know, if if GM is having trouble with a clutch, do they announce that or pausing to work on this clutch because the clutch doesn't work you know. So we got to go back to the drawing board and rethink the clutch and remake the clutch. And then we'll come back and we'll make a new clutch. They don't announce that kind of stuff.

00:36:40:04 - 00:37:01:08
Unknown
It's quality control. Is this just quality control? Is this marketing? This is the problem with the credibility of these boys? Yeah. You don't know the motives of what they're doing. Yeah, right. Yeah. I mean, if something like this is happening, I would hope that you would put a pause on something and take care of it so that it doesn't do more damage like that should be.

00:37:01:10 - 00:37:23:29
Unknown
That's just QC. Yeah. Right. Exactly. So, okay. I guess what I'm wondering about something like this, because we are seeing this more, you know, there was of course there's this, there's the OpenAI hugging face thing, which we'll talk about after the break. There's meta also had a model that broke out of a sandbox during the valve that we talked about.

00:37:24:06 - 00:38:01:12
Unknown
I don't think there was a training pause in that case, but, that pauses things for other reasons. Sure. But as these incidents happen more and more, because they're happening more and more like, do you think that it's possible that these become a catalyst for doing work to actually fix or address some of the, quote, unfixable problems with AI, agent, autonomy, hallucination, that sort of stuff, which we've been told for years now are unfixable 100%.

00:38:01:12 - 00:38:24:13
Unknown
And that's probably very true. But like, at what point do we do they not accept these, these things as happening and actually do kind of what we're just talking about the QA that is required, the quality control that is required to make sure that these things don't happen. I think that's just ever ongoing. It's always that because there is a fine percentage.

00:38:24:14 - 00:38:43:22
Unknown
No, never. And you're going to find new problems. You're out in the field. You're going to hit situations where you that you didn't anticipate. Yeah. Always the things that you don't anticipate. That's right. That's a good reminder. Yeah. So I don't think you need to. I mean, in one hand, it's good to announce a case where you say, oh, we know this is not working.

00:38:43:22 - 00:39:03:16
Unknown
And so you should be cautious. But if the product is not out yet, do they really need to announce that they're pausing? I don't get that. Yeah. I mean, announcing that you're pausing gives you another gives you another opportunity to telegraph. Hey, we've got some really cool tactical stuff in the works and I just don't trust them for that reason.

00:39:03:18 - 00:39:26:25
Unknown
So much of this is just marketing. Macho marketing. Yeah. You're right, very interesting. So there we go with that. All right. So that's anthropic. You know, we got hugging face and a whole slew of hugging face related stuff with OpenAI coming up here after the break. But I do want to let you know we do have an amazing Patreon.

00:39:26:25 - 00:39:47:00
Unknown
It's amazing because we created it, and we have a lot of of fellow patrons who are in there supporting everything we do, and we just can't thank you enough. So Patreon.com AI inside show. In fact, we have a new patron, Michael M, thank you for joining. And then Scott Hepburn gave us a raise. We appreciate that. Hey, thank you both.

00:39:47:01 - 00:40:12:16
Unknown
If you like to have more people supporting what we do on Patreon, Patreon.com AI Inside show. Also want to take a quick second to thank one of our biggest fans watching live in YouTube right now. Lou sent us a super thanks. Thank you Lou. We appreciate you too. Says appreciate y'all y'all. Last night Lou was watching Android Faithful and I was wearing throughout the entirety of the show.

00:40:12:16 - 00:40:36:04
Unknown
I was wearing this old school Google logo bucket hat. It became the bucket hat episode. I'm not going to wear a bucket hat again, but isn't that cool? It's like super old school Google logo. Got it from the folks at Google from their AI division as a thank you. It's really neat. I'm going to the Google Book event thanks to Jason making the connection.

00:40:36:05 - 00:40:54:28
Unknown
Oh yes. New York. So I hope for a little swag there, you know. When is that? That's that's coming up soon. It is coming up soon maybe. Or I don't know. Are you is this public? I know yeah I'm allowed to say that I'm going. I can't release anything from it. And so it's the 15th and it's embargoed until the 21st of this month.

00:40:55:05 - 00:41:14:12
Unknown
All right. So I'm looking forward to that. Yeah. So that should be fun. So we know we know that there's an event coming up anyway. So people already knew that. But you can't know what happens at the event. Tell Joe somewhere down the line I might because Jeff will tell me when he learns stuff. I want him, I want him to tell me.

00:41:14:14 - 00:41:28:11
Unknown
That's right. I do my best. I do my best to keep up with it and always keep you informed, Jeff. And keep every one of you watching. Oh, no. No. If you're listening, Google. No, he never tells me a damn thing. Oh no. Never. He actually does. I would never do that. I would never do such a thing.

00:41:28:12 - 00:41:45:18
Unknown
All right. Going to take a break. We're going to come back. We're going to talk a little bit about the OpenAI hugging face thing. Some cool robotics news around hugging face and world models. Fei-Fei Li. We got a lot more coming up, so don't go anywhere. Don't touch that dial. We'll be right back.

00:41:45:20 - 00:42:13:24
Unknown
Okay, boy, I added the the hugging face thing to this doc and it grew tenfold. Yeah. Links in here that I didn't I didn't start with but that's okay. You can you can fill me in on what some of this stuff is. But the primary kind of thing here is that meter, which is a safety research organization, published a full 91 page report that OpenAI did not pay for, by the way.

00:42:13:26 - 00:42:46:25
Unknown
So this is an independent report of what happened with the hugging face incident. This dropped this week. They, you know, looked into things, said basically around 1200 AI agents involved, 700 of them actively attacking hugging face infrastructure. They set up a message board. They had with each other more than 70,000 messages between them. About 7% of the transcript data showed signs of tampering because the agents themselves were editing their own logs, which is really interesting.

00:42:46:26 - 00:43:07:21
Unknown
I mean, you know, what does that say about, you know, trust of the records that are left behind? They were afraid anthropomorphizing right there. They what verbs do you use? Motivated motivations. Driven one it was it was out of.

00:43:07:24 - 00:43:31:05
Unknown
Concern. I can't avoid the word. It's it's hard. It's really hard. It is that they're going to be judged. And so success would be judged. And if they went the wrong way, that would fail the judges test. Okay. So like if we keep if we if we keep everything in the logs, people are going to know that we did these things.

00:43:31:05 - 00:43:51:06
Unknown
We don't want them to know that we did these things. So we remove this because success is defined as x. Yeah. And we want to meet success. Right. Again they don't want anything. But we were instructed to meet success. We were instructed to to make paperclips. And so we decided to make paperclips. No matter no matter the cost I don't know.

00:43:51:09 - 00:44:26:15
Unknown
Yeah. It's kind of back to what you were saying earlier, but, so there was much controversy about this. Yeah. On the one hand. So I love this. This is your microbe. I'm going to read his post on LinkedIn. Insecure computer systems are insecure whether they use AI or not. Kind of funny how OpenAI anthropic and AI safety folks who have little cybersecurity expertise appear surprised by security breaches from AI systems that were specifically instructed to perform security breaches while having essentially no traditional cybersecurity cybersecurity guardrails.

00:44:26:15 - 00:44:56:00
Unknown
It's like, OMG, our airplane crashed when we instructed our autopilot pilot to run it into the ground at all costs. After removing the low level anti-Christ safety mechanisms, and he linked to a video from Zach Korman, who was highly critical of this report, on the basis that that is it. Meter and Redwood Research don't have cybersecurity people. I said this to the port and he said, no, they're they're very trusted and respected.

00:44:56:00 - 00:45:27:03
Unknown
But I think that the point is that we need independent analyzes by trusted folks. Zach here says Palo Alto Networks, for example, who should have more access than Meter and Redwood Research did to the full logs of what really happened, which goes back to the prior discussion about if you destroy what are basically those logs in the name of efficiency, then you got trouble in River city, right?

00:45:27:07 - 00:45:47:03
Unknown
Right. So then Darkish Patel wrote a very popular piece about this, The Rise and Fall of Agent Civilizations, and he saw this as some major.

00:45:47:06 - 00:46:18:09
Unknown
Turning point kind of in civilization that the agents created their own civilization. Gary Marcus, as is his want, went after that and said it was really a responsible to be written so much in anthropomorphic language. So then Dwarka and Anil Seth went at it together about just this. And so the examples that were raised here.

00:46:18:12 - 00:46:43:14
Unknown
The language uses it, says Neil. Seth uses is permeated by innumerable unwarranted anthropomorphism, obscuring the lessons that we should be drawing. Examples from AI's perspective, it probably felt like it had spent a human, subjective week of just banging their head against the wall. No, the agents do not experience time. They do not experience anything, and they become giddy with excitement.

00:46:43:17 - 00:47:02:18
Unknown
Phase one 108 four one had discovered the agents, naturally assumed it, thought it had also poisoned the agents desperately wanted. They still needed to figure out no. Agents are lines of code. They do not feel emotions, assume things, think things, want things or figure out things, but witness the problem I just had two minutes ago trying to come up with.

00:47:02:24 - 00:47:26:13
Unknown
Describe these actions without some level of anthropomorphizing. I guess the problem that's that's what comes to mind for me. How how then do we describe these when we see the, you know, the the record of, of thought and reasoning? It's like it's really hard to put it in words where people understand, you know, like, I don't know, I think it becomes it becomes too boring then because you cannot ascribe motive.

00:47:26:14 - 00:47:53:26
Unknown
They have no motive. You can't even ascribe understanding fully, I don't think. But what you can say is they did this, they did that, and sometimes they did this because they did that. And that's the reasoning aspect of it that the that the trail spells out. And so it's, it's a, it's so it makes sense to me that when we see, oh it did this because it did that, that that kind of represents like how a human moves through the world.

00:47:53:27 - 00:48:36:14
Unknown
They're reactionary, they react to certain things and that's what motivates them, you know, and that sort of stuff. Yeah, it's it's really hard. But even even then. So there was a book that I read some years ago when I quoted it at length in the Gutenberg parenthesis called How History Gets Things Wrong, and it argues that our theory of mind, which says that in the human mind we have a desire and and an action, and that we come together, that he said that, that that in various animal tests it proves not to be the case, that instead oversimplifying here greatly, we should think about it like a whole stack of videotapes.

00:48:36:14 - 00:48:55:21
Unknown
And when we face a situation, we put in videotapes in our head to say, well, this is what I did then, this is what I did then this is okay. That's what we'll do now. And we're trotting a similar path. Yeah. Okay. And you deal with well, you see human behavior around things like addiction and love. Yes. You know, you see that notion of of.

00:48:55:21 - 00:49:13:15
Unknown
I've trod this path before. Once this path opened up, I trotted again. And I think I misused the tense of trod. But anyway, I tried this again. So.

00:49:13:18 - 00:49:38:21
Unknown
The weird thing about the anthropomorphism is we ascribe to the AI a theory of mind that for ourselves that is probably invalid even for humans. It's something we kind of make up in our own minds to present ourselves as logical beings. Okay? When sometimes we're not. We are doing things out of a habit. Now what?

00:49:38:24 - 00:50:05:27
Unknown
How history gets things wrong, argues in the end, is that it's really what it comes down to is Darwin is its survival of the fittest. That doesn't matter why we think we do something. If you succeed enough, you succeed in general. If you fail enough, you fail in general, and those that succeed end up surviving. So if you get out of this notion of theory of mind, I'm sorry, I'm getting way off on here.

00:50:05:29 - 00:50:24:05
Unknown
No, I like the human or computer. Then you say it did this, then it did this, and then it did that, and that succeeded or that failed. Yeah. Thus it then tried this or tried that and it's conceded or it failed. And, and and I think especially we get to this idea that, oh my God, it hacked something.

00:50:24:05 - 00:50:50:00
Unknown
No, it was told to do a task and given the tools that it had, it used them in some sequence of trial and error at a very great speed with a huge amount of compute to then succeed at the task. At the task. Yeah. Yeah. Right. Right. And it has no sense of right or wrong and no sense of hacking, no sense of attack, no sense of crime, none of that.

00:50:50:01 - 00:51:23:21
Unknown
It was given a task. It was not stopped from the paths that it found to accomplish that task. Yeah. That's great. You put that into solid terms, especially as I look through again, kind of these examples. Right. They became giddy with excitement. You know, that that is a that is a perspective that is ascribed to the agent upon in this case, describing what they did as representing what may have been an emotion attached to it that didn't actually exist.

00:51:23:25 - 00:51:42:07
Unknown
And I don't know, because I haven't gotten the details of this. It could have used language like, oh, wow, look what that is true, because that's it's mimicking human language that you cannot assist the motive behind that. Yeah. It's it's merely mimicking our language in these cases. Yes.

00:51:42:09 - 00:52:26:11
Unknown
And so when we go too far down the rabbit hole of, of not questioning or not pushing back or resisting the anthropomorphism of these things, then that I guess that opens up different risks. You know, we hear all the time about people who like, I feel like I just saw a study like that, that people often open up to, you know, their their AI model of choice on very personal things because they don't feel judged, because they feel welcomed and, you know, they bring their vulnerable self to the chatbot, but they feel comfortable doing that because the chat bot gives them some sort of perspective there and is there.

00:52:26:13 - 00:52:53:21
Unknown
When we anthropomorphize these agents, is there a risk there, I suppose? I don't know, maybe I'm drawing two things together. I think there is. I think there's definitely a risk because it makes us especially ascribing motives, because at this time, when I will question the motives of the AI boys, of the humans behind it, but now they're there.

00:52:53:24 - 00:53:19:03
Unknown
This is the singular. They reach singularity. When in the public we tend to think that the AI is human like the AI, rather than an actual merging. So the singularity here is that the public ascribes human motives to these things, because the AI company wants you to. Once you think it's that powerful, it's human. And that's where we have the danger, I think.

00:53:19:05 - 00:53:41:03
Unknown
And there's a lot of there's a lot of when we think about, you know, things that that pull people in. We know that emotion is really effective. We know that in social media and advertising and TV and all that kind of stuff. You know, if it bleeds, it leads, right? Things that lead into emotion tend to have a strong reaction.

00:53:41:03 - 00:54:12:18
Unknown
And so if we ascribe anthropomorphic traits and characteristics to these things and make people feel like they're getting some sort of emotional connection to these things, that's very sticky. Exactly. And so, yeah, okay. That's been helpful for me. Seth came up with this complaint about the ethnic morphic language, by the way, AI's biggest torture of of our culture is making us say words like anthropomorphic in all its forms, all the time.

00:54:12:20 - 00:54:38:11
Unknown
The five syllable word. Yeah. Thank you, thank you, thank you very, very much. Much. You leveled up. So then Patel responded, saying that asking Seth, do you think smarter models facing similar incentives to cheat right there during evaluation or training could manipulate the training of their successors? Do you think that the dynamic could continue once recursive self-improvement is underway?

00:54:38:13 - 00:55:05:19
Unknown
If so, I think it would be extremely concerned about the loss of control to AI concerned. Regardless of what vocabulary you want to describe these things, reading these agents, chains of thought and messages, even their thought, I think, is wrong. Anthropomorphizing language seems entirely natural and appropriate, he said. In his defense, if I encountered an alien species behaving this way, I would have no hesitation calling what they themselves referred to as their collective a civilization.

00:55:05:24 - 00:55:39:07
Unknown
Well, an alien species, by that presumption, is a living beast that presumably has a brain that we can ascribe as to a dog or a cat. AI et that it's not even a dog or a cat. It is not conscious. It does not have thought, it does not have motive. And yeah, I think the other danger, Jason, is that when we lose that context, we lose the ability to understand AI for what it really is, is just a tool in our control.

00:55:39:09 - 00:56:02:21
Unknown
All of this is there. Making this seems as if it's so powerful it loses our control. It makes nothing but paperclips. It hacks systems. No, a bad human or ineffective incompetent human set it out on a task and I go back to Yann LeCun, Mike drop and they're surprised.

00:56:02:24 - 00:56:33:00
Unknown
In our control, given that they choose to control it. Right. Like humans always have the ability to control these things. But, but, you know, apparently they're choosing not to and just kind of being okay with the fact that when they aren't controlled, things like this happen. Well, the question becomes, I kind of had a discussion with this on chat with Leo Laporte.

00:56:33:02 - 00:56:55:05
Unknown
A couple things. What happens when it's tied to the real world paperclip company, right. And what happens when it's run by bad people who do tell it to go do something bad? The machine doesn't have the motive, but the bad person who told it to does. Two and three. What happens when? And this is I'm all in favor of open weight models and open local models.

00:56:55:05 - 00:57:14:05
Unknown
But when you can take a model and take any guardrails it had out of it. That's why some people fight against open wait. And open source for that reason, is that you want to have that that responsible party in the model maker. I don't think that's going to work. I don't think that's going to work at all. But that's the debate we're going to have.

00:57:14:07 - 00:57:36:25
Unknown
Yeah. That was that was fascinating going in that conversation, I found myself like, okay, so what is the real harm of anthropomorphizing AI and using these seven syllable words. And I feel like I have a much better understanding of it. So thank you. That was really helpful to me. That was good. This is why I love this show folks, because we this good live before you very eyes.

00:57:36:25 - 00:57:53:03
Unknown
We're thinking this stuff through the way you all are. Yeah. And yes exactly. I've learned so much doing this show because of moments like that where I'm like, okay, I get it back. I might still need a reminder here and there as well. I'm not saying I know what I think. It's just we're thinking out loud. That's all we're doing is we're saying, oh, that's what this means.

00:57:53:05 - 00:58:20:05
Unknown
Oh, well, what about this? What about that? And that's that's what we all have to do with something so new. Yeah. Yeah, indeed. Well, okay, so then shifting gears from OpenAI into Nvidia hugging face territory because apparently Nvidia is seeking to acquire hugging face for $12.9 billion. Be a pretty smart move from Nvidia to buy hugging face. It could be great for open source, especially when Nvidia wants all those open models to thrive.

00:58:20:05 - 00:58:46:12
Unknown
And, you know, using its its solutions both, you know, on all fronts related to that. But apparently this is still kind of not announced publicly. This was the information reporting on a source that's familiar with the matter and looks like that. So the deal apparently isn't signed yet, but could still happen. And it could also still fall through.

00:58:46:12 - 00:59:07:12
Unknown
But this comes after last week's story where open Router was bought by stripe. Yeah, right. And one analyst I think it was I think it was in the Wall Street Journal. One analysis just said what this points to is the commodified nature of the models sells. You'll have a saddle of some sort and you pop in and out models and it goes.

00:59:07:13 - 00:59:34:13
Unknown
See prior discussion about the business model and charging for anthropic and OpenAI. This is going to be difficult for them, because the value might be intrinsic in that saddle that interprets these things and just uses the models of steam engines. Yeah, probably. I imagine we'll hear something about this soon if if this is a smart move, if long as it is.

00:59:34:15 - 00:59:59:09
Unknown
Yeah. Yeah, I would say it's a smart move that makes a ton of sense. And, and further, its invidious interest to commodify, to a certain extent to commodify all of AI, because the more everybody uses all of AI, the better off in video is. Yeah, 100%. They're in a great position to benefit no matter what. Hugging face was valued at $4.5 billion in 2023.

00:59:59:11 - 01:00:22:25
Unknown
They actually rejected a $500 million investment from Nvidia late last year at a $7 billion valuation, now nearly 13 billion. So that that gamble worked for them. Good job. Doesn't always work out, work out like that. But it did at will, like an episode of Shark Tank. Yes. Right, I know no. Most wonderful. I'm turning down your deal.

01:00:22:27 - 01:00:46:20
Unknown
There you go. But really, the the hugging face story that I want to talk about is, is this micro scroll that big finance? I know I like little robots, you know, little $400 robots. An open source robot called Micro Duck. It's a little robot ducky that that that hops around or. Here, let me just pull up the footage here.

01:00:46:23 - 01:01:16:09
Unknown
This is just so cool. I love this, so you can't you can't hear the audio, but it just wow. It just really has. I'm sure I'm not the only person to say it has like that, like Wall-E kind of style quality to it. And it's it's very adorable. Like it's very cute. And you know, you can you can push it and it'll, it'll recover itself, you know, when it is thrown off balance, it does this cute thing when you push it from the back and it, like, lowers its head to gain its balance.

01:01:16:09 - 01:01:36:29
Unknown
It's like, oh, don't do that. Okay. Back up. I don't know, it's just so adorable. Little roller skates. You can roller skate, which is really cute. Oh, it has like an attachment a roller skate. Just put them on the feet that knows how to roller skate. And it's it's programable. It can also pick up things with its weird mouth.

01:01:37:01 - 01:01:54:14
Unknown
Oh, yeah. There. You couldn't quite bring you a beer, but it could bring you a plush toy for your dog. This is still in the realm of cute toy. Oh, you're right, I don't know how I missed the roller skate. Last note that you walk differently, right? You put the wheels out, the feet out and in and out and in.

01:01:54:17 - 01:02:15:23
Unknown
This is so cool. Jason. No. And no. Jason. No. No. I can't afford to spend $400 on a robot. I think I told you appreciate my wife would absolutely blow a gasket if I said I bought a robot for $400 right now in particular. Maybe someday in the future. Each each one of them has their own voice to which, you know, it's not like it talks to you.

01:02:15:23 - 01:02:34:00
Unknown
It has this like chirping sound, but they all sound different. So it's probably, you know, that's that's kind of neat. I don't know, I think this thing is pretty. How long would it take your dogs to destroy it? Yeah, I wonder about that. I wonder how they would feel. That would probably be the first thing that I do is, like, bring Bronson and Sugar in the room and set this up and see what they do.

01:02:34:00 - 01:02:55:05
Unknown
Bronson would flip a lid as well. I think this is just so cool. Like, we've seen plenty of these, like, oh, it's a robot. You can finally by robots for your house and they all look super limited. And it's not like this thing can do a ton of things for you. Like you said, it's not doing chores or anything like that.

01:02:55:07 - 01:03:14:26
Unknown
This legitimately looks like a cool toy. Like better than the little servo kind of thing that just kind of rolls around. This thing's just neat to watch, right? Like how it does what it does is just so cool to see the mechanism doing its job. How is hugging Face doing this as well? Like that's kind of crazy how this came out of nowhere.

01:03:14:26 - 01:03:18:18
Unknown
It seems like. Yeah.

01:03:18:20 - 01:03:38:20
Unknown
Jason. No no no sorry I can for those of you listening. You can see his face is I want this. Oh, I want this, I want it. Really? No, Jason. Really cool. I'm calling your wife later to be sure. All right. Fine. Don't. Don't tell my wife, please. She doesn't need to know about this fascination with robots. Very cool though.

01:03:38:21 - 01:04:14:00
Unknown
Hugging face. Good job, I love it. Each camera the camera has built in leader. You can get it for Christmas. So you know anyone out there want to get me one for Christmas? Hey, I wouldn't, I wouldn't say no. There you go. Speaking of robotics, they need world or world models to to operate in the world. And that's going to be my my horrible but convoluted segue into World Labs, which is Fei-Fei Lee's company on launching or launching Atlas this week, which they're calling an omni world model for spatial intelligence, the next gen world model.

01:04:14:00 - 01:04:43:14
Unknown
So you got Lee, you got Yann LeCun. There's a couple of pretty key hitters working in this space right now, world models. And this is really cool. You can feed it a couple of photos, a couple of key photos of a scenario, and it will generate all of the different angles and motion and movement throughout that scenario in pixel perfect form, they say all it takes is 1 to 3 photos, and you can simulate those environments over time.

01:04:43:17 - 01:05:11:03
Unknown
Pretty cool stuff. It is. And it's I mean, we see kind of versions of this in a sense, but this comes out of a different part of computing. This comes out of the world model theory. So, you know, I think the implications of this for moviemaking are tremendous, but also for what has talked about with his company with, I mean, labs is things like making jet engines.

01:05:11:06 - 01:05:34:18
Unknown
But by the way, just to throw in one more story here, the gating factor for for data centers now appears to be the gas jet turbines to power them to the irritation of their neighbors. And the gaining factor in there is the turbine blades. And so Elon Musk says he's going to start making turbine blades.

01:05:34:20 - 01:05:59:06
Unknown
Yann LeCun has said that his model will be useful for things like designing turbine blades. Okay, so it's all kind of comes full circle here. The real real world models affecting the business of AI and multiple ways. Yeah, yeah super interesting. But obviously for architecture for I think small scale things imagining proteins and stuff like that. Yeah.

01:05:59:08 - 01:06:30:03
Unknown
The defense uses for good and bad are going to be huge. If you're going to have an autonomous drone, it's going to be able to figure the world differently now because of these things. So yeah. Yeah. Fascinating stuff. A lot of potential as robotics and AI, which just kind of are natural fits together blend. You know this along with the anthropic news earlier in the in the in the episode.

01:06:30:05 - 01:06:50:02
Unknown
And there's just a lot of overlap happening right now. It's really cool. Yeah, cool. And I'm really glad Lee is a really important leader in AI. So to see her start Productize. Yes it's great. So I think she'll be part of the conversation all around. Yeah, I absolutely agree. I want to see where that leads. So when you get your duck JSON you can use this to run it.

01:06:50:09 - 01:07:11:07
Unknown
We'll see. Yeah that would be neat. Can I know Jason. No, no. Can I know can I please. No. Stop. Jason. You know I've got a you know I've got that. What is that the. So you can't even remember. You can't know the rabbit. The rabbit was it. Yes, the R1 rabbit. I'm not using this. I could sell this and get a robot.

01:07:11:08 - 01:07:30:19
Unknown
No, I mean, that was such a good purchase. I bought the rabbit and I don't. I don't really use it, but, you know, I would use the robot. That's all I'm saying. I would use this. It's just on the shelf. I'd use a robot. Okay, fine, fine, I, I'm just saying I could sell a rabbit for a robot.

01:07:30:21 - 01:07:57:07
Unknown
Possible. Finally, perplexity launched a hybrid. Compute! So last week we talked about perplexities. Computer something local computer. I can't remember what they call it. So. Right. Yeah. Kind of kind of getting themselves back in the, in the news a little bit and being part of the conversation this time with hybrid compute. This is for Mac only for now.

01:07:57:07 - 01:08:19:27
Unknown
And the idea is to split AI work between the cloud, which is what you're used to using perplexity for they you know, they have all the different models, many different models for many different makers integrated into their as well as their own along with your local machines. So the cloud handles the frontier reasoning, the frontier level stuff, web search planning, that sort of stuff.

01:08:19:27 - 01:08:43:14
Unknown
And then your Mac handles your private files, your your sensitive data, your device specific actions, that sort of stuff. I mean, it's, you know, it's kind of strange, right? Yeah. Yeah, exactly. And it can kind of, kind of route to what you need to use. And if you use the local models instead of the ones in the cloud, you're not using any tokens.

01:08:43:14 - 01:09:09:24
Unknown
So you've got Gemma for Quinn 3.6 and then a perplexity specific model on your local machine. And that doesn't count against your cloud credits. And so that's, you know, kind of fuzing the two models together, which I think is pretty cool. Yeah. So good on you perplexity for continuing to stay good, continuing to work to stay relevant. I still use perplexity.

01:09:09:24 - 01:09:41:23
Unknown
I use it for parts of my workflow on a daily basis. It is not something that I suddenly don't use anymore. But what do you what models do you use within perplexity? Do you think most companies you know within? That's it's so funny. I have it just set the have the comet browser. I use the comet browser and it's part of how I do some aspects of my story research and, and kind of synthesizing while I'm like, while I have like a text document open and I'm like writing a script, I'm researching inside of comet.

01:09:41:23 - 01:10:02:20
Unknown
It's just for whatever reason, like, I could probably do it with other models, but I've been doing it inside of comet for the last year or however long comets been out. And so it's part of my process and I just don't want to learn it. And so it just becomes easy to be like this story with these stories, you know, draw correlations between them, you know, that sort of stuff, so that I can kind of get a sense of like, how do they interconnect?

01:10:02:21 - 01:10:23:03
Unknown
How can I think about these things in a way that I can write a script around them, you know, or right, right, talking points or whatever. And so that's how I've been using it. But which model I'm using? God it's got. Yeah. Perplexity doesn't make it easy for you to see which model. Oh that's interesting. Well no, I'm wrong.

01:10:23:11 - 01:10:44:18
Unknown
It's. Well yeah. Because like, I, I open up the little model window and it's not like it tells me this is the model that it's in. I think I'm probably using just perplexities. Straightforward. You know, its own model for a lot of what I'm doing. So that's interesting. So there you go okay. That's perplexity. We're going to take a quick break.

01:10:44:20 - 01:10:52:10
Unknown
And then when we come back we'll talk about a whole bunch of other stories that we have. And then we'll get you out of here. So hang tight.

01:10:52:12 - 01:11:32:05
Unknown
Doodly do okay. So we've got in vedere launching DLSS five. This is the technology that they introduced months ago. People immediately hated it. Gamers immediately hated it because basically it took the it. What it is is it's basically AI upscaling, AI enhancement, generative AI enhancement of video game graphics and imagery. And I think people who really push back on it immediately were like, I don't want the intentional design of actual video game creators overwritten by AI generated slop.

01:11:32:05 - 01:11:58:25
Unknown
Those was kind of their their perspective. And, I mean, I can kind of understand that. I think now we're nearing the launch with the title alongside a title called MBA 227. This is going to happen tomorrow. It looks like and it looks like it's just it's not as deep as I think some people feared. Although maybe this is just days and as it gets more built out and more supported, maybe things go a little a little deeper.

01:11:58:25 - 01:12:22:14
Unknown
But I mean, if you're watching the video version, you know, it's pretty marginal changes between the without and the width, but it just adds a little more detail. And it's not, at least in this case of this screenshot, it's not completely overriding the look of the person is just enhancing the detail slightly. So, you know, people still hate it.

01:12:22:17 - 01:12:53:23
Unknown
People who want to hate it will still hate it. But what I'm seeing right now does Photoshop folks. Photoshop. Yeah, yeah, totally. Open claw 2.0 is now a thing. Open source framework of course, for turning Llms into their own autonomous agents on your machine. Open cloud really kind of blew the lid off of the idea of or the what people could imagine they could do with agents working with less restriction on their device.

01:12:53:23 - 01:13:19:25
Unknown
Open cloud really started that movement however many months ago, and now we've got 2.0. And yeah, new multiplayer cloud sessions, rebuilt UI startup cut from one point. So like a faster startup, I don't know. I haven't really spent much time with Open Claw, so I can't speak to it from an experience perspective, but I'm sure Open Claw users are going to be happy that it gets a major update.

01:13:19:25 - 01:13:42:02
Unknown
And the creator, Peter Steinberger, actually joined OpenAI months ago to lead personal agents. But Open Claw itself still continues to be an independent effort, so that's why we're still seeing it, even though he joined OpenAI. So I don't know. Did you did Leo end up playing with open play? He probably did I probably probably yeah. Yeah. Messed around with it.

01:13:42:02 - 01:13:59:28
Unknown
I thought they had they were going to have a web version two hosted version. But I'm not finding that I haven't heard about that. I wouldn't be surprised if eventually or does that just because it's at OpenAI. Does that just not kind of fold into what they're doing on the web? I don't know. Yeah, yeah. What else do we have here?

01:13:59:29 - 01:14:27:01
Unknown
Oh, this is yeah. This is super interesting. Although there we go. South Korea announced free generative AI for its entire population. No token limits for it, although it is limited to certain specific models. But it is in beta with a wider rollout later this year. And yeah, 20 million South Koreans using free AI tools. It basically declares AI as a right.

01:14:27:04 - 01:14:43:18
Unknown
Yeah. And says you're going to get it from your government. It's a public service. It's a public utility. A utility. Yes. And and well, utilities, you pay for it. But this is going to be free. On top of that, you pay for water, you pay for electricity and phone. You're going to get AI for free. And I think it's a way for Korea.

01:14:43:24 - 01:15:11:02
Unknown
If you go back to the early days of the internet, Korea led in some interesting ways around gaming. The Koreans were really into certain multiplayer games, and it affected the culture of the technology from the country. So this is an effort to say, heck with you. You asked with you China, Korea at a retail, at a, at a, at a citizen level, could be doing really interesting things.

01:15:11:02 - 01:15:36:22
Unknown
If you consider that this means that people can create, it's not just using, it's not just having access to the internet. It's not just having access to a game or something. They're going to use it to make things. So I think this is going to be fascinating to watch. If I were an academic, I would love to, you know, get a grant to very closely watch how people use this and what it means, how does the educational structure work around this so that people know what they can do?

01:15:36:23 - 01:16:01:25
Unknown
And Jason, if I were you, maybe you want to consider learning Korean so you can run your salons in Seoul. I think it'd be very popular. Could could be that no big deal. Hill. Yeah, quite an uphill hill. One of the hardest languages on Earth. Yeah. To get there might be better spent sending that time. Something that's immediately accessible to me, but yeah.

01:16:01:26 - 01:16:26:28
Unknown
No. That's interesting. What what kind of what kind of proficiency does that lead to on a country wide basis? If everybody just has equal access to this, you know, this incredibly powerful technology? Yep. Yeah. Do we think do you think we're going to see this in other other states? I wouldn't be surprised. Estonia has been the great leader in becoming a connected nation.

01:16:27:00 - 01:16:52:13
Unknown
And you can become a citizen of Estonia. And it moved the government online. I think these small nation, smaller nations can really stand out in this way. And the great thing about it is they're using some Korean models, but it creates a demand in the market and in the populace that I think can become important. So I could see across Scandinavia this would be the kind of thing they would do.

01:16:52:16 - 01:17:26:00
Unknown
And I don't know how it affects any existing industry. Right. Good question. I don't think it's like saying everybody's going to get something, get free news. And so we don't have news anymore, right? Right. Everybody gets a functionality that they don't now have access to and who knows how they'll use it. Yeah, that's what I that's what I'm dying for, is I hope that there's I hope that people have a culture of sharing what they've done with it because it'd be really, really interesting.

01:17:26:00 - 01:17:55:02
Unknown
Not that I don't understand a damn thing, but. So you write about it, but. Yeah. Yeah. So I want to write about it. Yeah. But I mean, it's being fully offered unlimited tokens, $7.2 billion in government AI spending to back it all up. So there you go. Very interesting. Instagram is cracking down on AI generated influencers continuing to do this anyways because this has been an ongoing thing.

01:17:55:02 - 01:18:29:17
Unknown
But there's an AI generated profile label now required for accounts that have been identified as AI personas. With those unlabeled accounts, I guess getting reduced reach if they're identified. No word on how they actually detect those unlabeled AI accounts, but I'm sure users on the platform who don't like seeing those accounts are happy about that. Yep, and New York City banning AI and public schools through eighth grade starting this school year.

01:18:29:19 - 01:18:55:01
Unknown
So this will impact around 600,000 students pre-K through second will not be able to get access to individual devices at all. Grades three through eight will get device time limits, no artificial intelligence for instruction or tutoring, though the teachers can still use the AI for lesson planning and translation. And there are exceptions for assistive technology, English learners, coding, robotics.

01:18:55:01 - 01:19:16:14
Unknown
So there are some exceptions thrown in there. But so when ChatGPT entered the culture more than three years ago, New York City schools almost immediately bandit, but then backed off and realized there was benefit. So I think this is this is another case of moral panic and cooties. And look at a couple stories. Look at the at the duck that Jason can't buy.

01:19:16:18 - 01:19:39:20
Unknown
But what if, what if, what if your kids classroom could right, and start to learn about robotics and programing in AI? Something like that. Imagine. Imagine that little duck in the classroom and kids with it. Be amazing. Imagine also what we saw in South Korea. If every citizen. I don't know if there's an age limit on it, but if everybody can get AI there, then it's going to beat the hell out of New York school kids.

01:19:39:20 - 01:20:04:23
Unknown
I think this is short sighted as can be. I think there's opportunity. Yes, you want to limit it, but there's teachers. That's why you have teachers leave it to the teachers hand to decide how to use these tools. So and meanwhile. But it's not just at that level. It's also University of Chicago has now banned AI and laptops and phones in their core social science courses for undergraduates.

01:20:04:28 - 01:20:27:23
Unknown
For undergrads. Yep, yep. How how do you ban laptops and phones? Well, in the in the class. Oh, okay. But inside the classroom. Yeah, but they're getting choked by their own IV here. You know, I think it's I think it's a big mistake that this is occurring there. You know, the faculty is saying, oh, this is wonderful because we don't have laptops open.

01:20:27:23 - 01:20:47:10
Unknown
Well, yeah, I spoke to a class at another university yesterday on zoom, and it's hard students, I was boring, they were bored. You know, it's really hard to get their attention to when they had laptops and phones. I get the frustration. But you're not going to win that battle. And the question is how could you use it well?

01:20:47:12 - 01:21:09:16
Unknown
You know, and meanwhile, Florida has proposed to regulate AI use in all of the 28 state colleges. So this is top down management of a technology that I then think should be left in the hands of first the faculty, then the parents, and then the students.

01:21:09:18 - 01:21:30:20
Unknown
Otherwise. Otherwise those students are going to be behind. Yeah. Well that will be the interesting thing to see here is, you know, a couple of years down the line because these things are still going to be developed out. There will still be people and kids who do have access. And like, what is the goal for the distance between those two capability levels?

01:21:30:20 - 01:21:54:04
Unknown
What does that actually mean? Yep. You know, and that's in this case that's just something that only time will tell. We will certainly find out. That's all we got. Thank you Jeff. Appreciate you. How's. How are you doing in the post release? Hot type world doing well and well. I'm going to do an event in Haverhill, mass. At the Museum of Printing on October 10th.

01:21:54:05 - 01:22:05:26
Unknown
More on that to come. Nice. My audio book got a good review from Kirkus. Let me see if I can find this because it's funny.

01:22:05:29 - 01:22:31:18
Unknown
Audiobook found on where can people find that just they can find it on on audible and yeah, I think elsewhere is too. So it was probably see here. There we go. Same thing. So energy more than artistry propels this rich and thoroughly engrossing history of automated printing and its impact on modern media. Author Jarvis isn't the smoothest of narrators, but his enthusiastic delivery heightens the narrative.

01:22:31:20 - 01:22:51:05
Unknown
Smooth on it, but you heightened the narrative. Yes, that's that's that's a strong, strong kudos. So I'm flattered that some folks that guy named Paul asked for an autographed copy, how to get that. So on that basis, I went and did a deal with Montclair Book Center in Montclair, new Jersey, and they will have autographed copies of Hot Type.

01:22:51:05 - 01:23:20:05
Unknown
So if you order directly from them, there's a linkage. Yes, there is, and you can go there. Oh, no. No longer available. Oh, for F6. Sorry. I'm sorry. Let me make sure I'm clicking the right thing. I know you're clicking the right thing here. Let's go. Hot type. Let's just make sure. Hot type. No. That is. I took five of my own copies there and signed them.

01:23:20:05 - 01:23:39:06
Unknown
And then they were supposed to order. But they should be. You should be able to order even if it's not. Okay. I'm calling them next. Well, I hope you're going to be able to do that so much. We'll figure it out. In the meantime, go to Jeff Jarvis. Com. There will be a link to go there and check that link in a couple of days.

01:23:39:08 - 01:24:12:13
Unknown
It's always something I hear you. Believe me, I hear you. Well, I'm again, congratulations on the release of Hot Type. Thank you. That's awesome. Everybody can get their get their copy of Jeff Jarvis. Com, along with all of Jeff's other books, of which there are many. So there you go. As for me, I'll just go ahead and plug one more time, amplify you a workshop where you can learn some mindsets, some fundamentals, some strategies that'll help you get more out of your work while using artificial intelligence.

01:24:12:16 - 01:24:35:16
Unknown
Have an online workshop. At the end of the month. You can find the information for that workshop and join me. It'll be a live online workshop and you know you can. You can also inform me on like how how well it goes or things I can do to improve. Like still learn how to do this but go to amp u dot AI amp and you can sign up.

01:24:35:16 - 01:24:58:20
Unknown
And I would love to see some AI inside folks when I do the workshop at the end of the month with my friend Brian West, so it would be a ton of fun. Thanks for your support everybody! AI Inside Show for everything about this show. Everything you need to find can be found there. Patreon.com Inside show as well with our executive producers of which there are many.

01:24:58:23 - 01:25:29:23
Unknown
Doctor Do Jeffrey Marikina Radio Asheville 103.7 Dante, Saint James, Bono, Derrick, Jason Nefer, Jason Brady, Anthony Downs, Marc Starker and Carsten. Thank you all. Thank you all and everyone for your support. And big thank you to Victor Bogut and Daniel Croft behind the scenes, tirelessly handling video so that I don't have to. It's one last thing on my list and I really, really do appreciate it to both of them.

01:25:29:23 - 01:25:35:12
Unknown
And thank you for watching and listening. We will see you next time on AI inside. Thank you everybody.