Move Fast, Break Things, on Rocket Fuel

This is AI Inside, episode 55, recorded Wednesday, 02/12/2025. Move fast, break things on Rocket Fuel. This episode of AI Inside is made possible by our wonderful patrons at patreon.com/aiinsideshow. If you like what you hear, head on over and support us directly, and thank you for making independent podcasting possible. What's going on, everybody?

Welcome to another episode of AI Inside the Show where we take a look at the technology that is interlaced with that's a that's a throwback term, with, artificial intelligence in all directions. Join joining me, Jason Howell, is Jeff Jarvis. How you doing, Jeff? Hey there, boss. Yeah.

I've got bad FOMO. I wish we were doing this from Paris if we got invited to the AI summit. But no. No. No.

Not this time. Right. Next year. Next year. Maybe maybe so.

I don't know. Like, they show the, the room where everybody was in. I don't I it seems kind of like a small event. I think this was a It's big, but it's small. You know?

Yeah. It was small high level. I went to a thing called e g eight some years ago, which was all Internet and Sarkozy. And, it was huge, and the tweezeries and a big tent, it was quite the thing. I think this is a more elite small group.

Yeah. I'm not in it. Definitely. Definitely has the elite, elite factor going forward. We're definitely gonna talk all about that, here coming up.

But before we jump in, wanna thank folks for your support on Patreon. We just we really appreciate those of you who are, are joining us. You know, we're we're kind of increasing the number of, of patrons and it really helps the health of the show. Patreon.com/ai inside show is how you can join the Patreon like Ciprian Nicolescu. I hope I got that right, Ciprian.

But anyways, it's it's a pleasure having you on board. Thank you for being with us. Actually, you've been on board for a while now. And, yeah. We got more shout outs to to throw around, including a new executive producer that we'll announce at the end of the show.

You can do this patreon.com/aiinsideshow and, become a part of the community there. And, you know, of course, just go to AIinside.show, which is just the, the the page for the show. It has all the information that you need. You don't need to support us on Patreon in order to get the show, obviously. But we appreciate if you do that.

So I think that's about it. Why don't we dive right in? And, you know, you mentioned you mentioned at the top the the two day AI action summit action summit. That's kinda sounds like it sounds like it's gonna be a short sighted movie. Yeah.

Totally. 60 countries signing the final statement at the end of the two day Jason, was that was that all the countries who were there? No. It definitely was not. No.

It wasn't. A couple of, of of of countries that did not sign the statement. And by the way, this is a statement that pledges inclusiveness, openness, safety, and the ethical development of None of which should be hard to agree to, you think? I wouldn't think Rather, Anadyne. But Gee, which which were which were the jerky company countries who didn't sign it, Jason?

You'd you'd never guess. The United States and The UK did not sign the agreement. And The US these days, I Yeah. Could predict. The UK, I don't understand why they didn't didn't.

I don't I don't I don't really get that. Well, it's it's just this French thing. It had something to do with two or at least the reasons that they mentioned has to do with restrictive language, and, I don't know, you know, how you parse that out. What would what would have what would it have said differently that would have had them, agree to it? I don't know.

It kinda seems like from the from The US perspective anyways, there what's in vogue in the current administration seems to be anything that's kind of like a team, you know, a a team player type thing. We're not interested because we're doing what's right for us, and so we aren't gonna sign. That's that's my interpretation. I don't know what you what do you think? Yeah.

I I think you're right. Right. It's and what's interesting about this summit, I didn't follow the summit in detail because I wasn't invited. But what I what I read about the coverage, I saw one summary because there were two prior summits. The last one I think was the last one was in England.

And it was all doomer safety, oh, time. And from what I understand, this one was much more optimistic. And and and France is not known for its technological optimism, so that's a that's a good thing. And was, I think, far more constructive and trying to say, what do we have here? What do we need to do?

What do we need to pay attention to? This is the right way to approach it. So I I think it sounds like it was a, all in all, a fairly good meeting, and, just a a positive way to say this is here whether, you know, it's not going to say it a little bit, but it's not going anywhere. Right. And so how do we, how do we work smartly with it?

And I think that's the right attitude. Well, and you mentioned optimistic and I think the there's, there's optimism in the world of AI, which which this conference or this this summit certainly had. But then there's kind of like the next step of optimism, which is we can do no wrong. It translated, you know, president, JD Vance really urged the EU to prioritize growth over safety. So there's, like, optimism there's, like, healthy optimism, which is, like, you know, it's it'd be cool to kinda, you know, work with us, and it's not going anywhere, so why don't we embrace it?

And then there is yeah. And I I think the extension of that is remove all regulate regulations. Let it be what it's going to be. Let it run wild because that's how optimistic we are about this technology. We yeah.

And I think that's where, like, JD Vance came from. He said, quote, we believe the excessive regulation of the AI sector could kill a transformative industry just as it's taking off, and we'll make every effort to encourage pro growth AI policies. So really going kind of like, I guess I guess, away from the regulatory kind of perspective that the previous administration was certainly, cautioning towards. Yeah. And I think, Eric Schmidt also spoke to them, and and I think, I'm trying to call that up right now.

He I I was glad to hear that that he said, that we have to focus on open source. And that the the real lesson of DeepSeek is that you need development in a lot of quarters at the same time, and the way to do that is open source. And Europe, had been talking the EU had been talking about trying to forbid it. I think that's long gone. But, I also believe we've talked on the show many times that open source is the way to go here.

And I think that, that voice was heard loud and clear in Paris. Mhmm. Mhmm. European Commission president Ursula von der Leyen was talking about building AI adhering to, the EU's strengths. EU is gonna invest €200,000,000,000 in artificial intelligence, to catch up and, you know, could be interpreted as a way to catch up to the progress that's being made in The US, made in China, you know, talking deep seek, as you just mentioned.

And you you put in the article from Kevin Roose from the New York Times Mhmm. Where he summarized, in his mind five big takeaways from the AI summit. And one of his takeaways was that Europe seems to be having regulation regrets. And that's kind of what you were saying as far as France was concerned. Yes.

Yeah. It's really interesting because because because Europe in general and Germany in particular and France as well have prided themselves on being the leaders in regulation. And the world needs somebody to think about what could go wrong in a sensible way and and deal with that. That's fine. But that doesn't get them very far when it comes to, investment and development.

And so, yeah, I think that they're trying to say, will it best? Will it best? Yeah. Yeah. It seems what what came to mind for me around this, and there was some discussion in, in the discord, which if you are a patron, you get access to a a pretty fun discord.

We're talking a lot more about AI in that discord. And something that kind of occurred to me through, like, so many different things happening in AI and technology in general is that this idea of we've gotta keep these we've gotta, you know, have safety at the forefront, ethical considerations, all of these things. It almost feels like very suddenly they've fallen out of fashion. And I don't know. Like, fashion feels like the right word for me Mhmm.

When it comes to big tech. They're really making changes very suddenly to say, those old efforts, that's no longer in vogue. Now it's, you know, dereg it's removing the the regulatory hurdles. It's removing, you know, the language that might protect this person or that person or whatever. And let's just go for it and see what happens.

And that seems Yeah. It's it's it's move fast and break things is back. Yeah. Right. It's back.

Yes. But the tools to break things with are more powerful than they used to be. Uh-huh. Yes. That is so, so true.

It's different than it was. Few other highlights before we move on. Smaller countries now feeling like they too can be a part of the AI game thanks to DeepSeek as as one example. And and open source from from both of them. Source.

Yeah. Meta and Mistral. Mhmm. That's right. And then policy, policymakers being fed the idea that AGI is imminent.

And, and I think it was Altman that was, assuring Trump that in his mind, AGI was something like this is this is applying pressure on the administration to take AI serious is the fact that in your term, AGI is going to happen, president Trump. And You may not get to Mars, but you'll get AGI. You'll get AGI. And what does that mean? And you wanna be with history here.

Right? So let's work together. Yeah. So anyways, so that is the AI Action Summit in Paris. And then I'd say another, you know, kind of one of the big headline moments in AI this week has been, once again, the Elon Musk, Sam Altman drama chapter 20.

I don't know. It's my last count. The never ending soap opera. The never ending story. That's right.

That Elon Musk's x AI company, along with a bunch of investment funds, submitted. And I don't know know if they actually submitted at this point because OpenAI's director Larry Summers said he hasn't received a formal bid yet, but, an unsolicited bid to merge OpenAI's nonprofit assets with XAI to the tune of $97,600,000,000. Really easy to look at this immediately. Like, the the the vibes I got immediately were, oh, I remember when Elon Musk did something similar to this, you know, went on Twitter and said, well, I'll just buy Twitter. And that became For sure, by the way, about half the price of all of OpenAI.

Yeah. Right. I I I Which which is worth more, really? Oh, I mean, Altman Altman even punctuated this. Right?

His response to Elon Musk's tweet was no thank you, but we will buy Twitter for $9,740,000,000 if you want. Yeah. A lot less, let's say. 10% the, the total of of what, what Musk had said about, OpenAI. So, I I take pride that, this is a little my I told you so here that I've been saying for ages that ever since Altman said that he wanted to go for profit, that, and even before that, as Microsoft, was re adjusting its relationship with OpenAI, I've said that this cap table is going to be a bear to set that this is about who owns what piece of OpenAI is a for profit company who owns what assets.

This is all extremely complicated. It reopens every negotiation the company has ever been in with every investor, including and especially Microsoft. And so what Musk did by all accounts is really just throw he trolled. Yeah. He threw a grenade.

He threw a grenade in the room and said, I'm gonna overvalue, this stuff so that it messes you up and try to negotiate what gets left in the for profit and the not for profit. And all of that has to be approved by the state of California. I think it's California. Maybe it's Delaware, but it has to be approved by regulators. And so they will listen to valuations that have been set in the marketplace.

So but I I I think the the the immediate assumption is Musk is not buying OpenAI. But let's not make the mistake we made with Twitter. Like, he's not really gonna buy Twitter. Exactly. Sometimes the it becomes a self fulfilling prophecy to a certain degree.

And if that if that was compelled to happen, if, you know, let's say they submitted a a formal, a formal offer, you know, the OpenAI board has to consider it, right, the fiduciary responsibility and all that. And if it was decided, like, okay. Well, then we I mean, it really is kind of to a to a certain degree anyways. I realize it's not the same, entirely, but it feels very similar to kind of the Twitter thing and how that went down. I was like, no.

That wouldn't happen, would it? Yeah. And then he got stuck. Oh my god. That's happening.

Well, at the time we said he got stuck with it, he didn't want it. And I think that was probably true at the time, though though now, of course, he's worth many multiples more than the amount he spent, and he's standing in the White House with his kid on his shoulder picking his nose, docking the president and and running through every possible department of government because he had this. So, you know, I eat crow about the wisdom of buying Twitter. Oh, yeah. It it turned out to be very wise for what he wanted to do, and he had more money.

However, he's not paying attention to Tesla. Tesla sales are down 60% in Germany. Stock is going down in Tesla. That has an impact on him. We'll see where that goes.

But so Musk is unpredictable. We know that he could go farther. He could end up with OpenAI. I saw somewhere. I can't find it right now.

I think it was the Wall Street Journal asking, well, who's better to run OpenAI, Musk or Altman? But I think this was a game. I I I think this time it's a game. Seems like it. It's a game that if it were to become reality, I don't think Musk would have a problem with it.

You know? Even even if it was a trollish game, he'd he'd be very happy with the outcome. He, which I guess is is to say he had very little to lose to to throw this grenade into the room. And No. You said and and I missed this part, Jason, but you said it earlier.

So he's bidding for what part of OpenAI? The, to merge OpenAI's nonprofit assets. Yes. Okay. But I don't think that's even set as to what is gonna be for profit asset, what is gonna be not for profit asset.

Right. No. It's so part of the negotiation. It's totally, it's part of the negotiation and there's still so much uncertainty around how that, how that transition is even going to work out. This just complicates it even further.

Yeah. Which is what he wants to do. Exactly. He is the great complicator. Yep.

So what do we fucking watch? I don't think it means much more than that, but it, it's the never ending, soap opera there. And I don't know if I put this in the rundown. Did I put this in that? Oh, yeah.

Just, just for mention. So already the New York times has there's a play, called Doomers, which is about, the peril and promise of AI. It is based on the 02/2023 ouster of Altman from OpenAI. So someone has already turned this into entertainment. Right?

And Yeah. So, it's it's off Broadway. I don't know, what it really is. It's, it's a review in the times. I I I've no idea if it's any good or not.

I don't know if it's even worth going to see, but it's amusing that this is seen already as a soap opera and being turned into one. And, so we thought that a nerdy now soap opera that Yes. A year and and a few months later Yep. We've got a a a play of some sort that in the New York Times about it. Yeah.

Oh, Jesus. I just I just picked up one root paragraph from this. I'm sorry. What I gotta say but the characters erupt in flagrantly unlikely monologues as when Alina, the company's scrupulous chief safety officer, tells her colleagues about disturbing recurring dreams, one of them vividly sexual. Oh.

This is not the nerd world I knew. That wasn't in the original story. No. An interpretation maybe. There are mentions throughout of Elon, no last name given, but there doesn't need to be.

Seth annoyed with Alina Snipes. You should've just had Elon's baby when he wanted to. Oh my goodness. Sounds like such a soap opera. How can people hold out their heads in any pride in this?

I mean, anyway, it's it because it's all it's all a carnival. It's all it's all Yeah. Amusement, and it's huge money or involved. And jeez. And yeah.

So I can't wait for the sequel, which is about, Elon's purchase of OpenAI. It'll yeah. Get roped in there. It'll be an an ongoing saga. I mean, I remember the time when movies and plays and stuff like this, like, you kinda had to wait like a decade, a decade and a half.

You had to get some distance from it. Ten, fifteen years before anyone was like, okay. We'll we'll approach that. And now it's like, oh, it happened a a a year ago. We'll go ahead and write a a play about it.

K. Entertainment. Yep. People pay for it, apparently. Alright.

We're gonna take a quick break. And then when we come back, I'd say this is a this I'm super curious to hear what you have to say about the, the Thompson Reuters copyright lawsuit that's happening right now. That's a really big deal as well. That's coming up in a second. Alright.

We're starting to get some rulings on some of the earlier copyright lawsuits related to AI, and this lawsuit in particular was first filed in 2020. So, you know, from quite a ways ways down the line. Thompson Reuters just won, I think I'm saying this right, just won its lawsuit against AI startup, Ross Intelligence, which is a, a legal research or was sorry. They're no longer around, but was a legal research company. The ruling says the startup was not permitted to copy the editorial content of Thomson Reuters to build its AI based legal platform.

It did not constitute fair use in this case. Like I said, ROS Intelligence is not around anymore. They actually shut down back in December 2020. So Because of the suit. Because of the expensive suit.

Because of the suit. Mhmm. But here we are five years later, finally kinda getting the the judgment there. The judge had initially concluded differently, and I think my understanding of this is a little fuzzy. He had he had concluded sometime last year the and ultimately denied the motion for summary judgment, but then reconsidered the judgment opinion, which is what led to this week's announcement.

So I'm not sure exactly where along that time frame the judge decided to reconsider it or why, what, you know, what the detail is there. But nonetheless, here we are. Yeah. And and this is this is a complicated all cases are complicated. It's all cases are unique.

But this one, I I don't know what the precedent value is of this. I'm concerned about the rejection of fair use because I have been arguing that that, training models on fair use, is fair use and transformative. However, in this Ross case, there's a couple of factors. One is that Ross used it to build its own new product that did depend heavily on what came from Westlaw, which is a which is a Thompson Reuters product, a. And b, this whole area of law and copyright, is very complex because laws are ours.

Right? That that it's our our government and our, structure, and those are both those belong to us. There are cases going way back, into the mid eighteen I've I've I've just wrote about one of these cases in the book I'm writing right now, where, a stenographer used, the transcripts of courts to create a new product was very similar. And in this case, the stenographer won. It was taken as fair use at the time, and the belief was you really couldn't copyright the law.

But that's changed a lot. And what happens is that you can in the sense that Westlaw has its, classification classification, interpretation, organization, taxonomy, Got it. Whatever, on top of the law and that that's what they own and that Ross benefited from that versus the law. So it's it's a little bit different from from, I think, saying, well, we read open web pages Right. And use that to train.

They did more than train. They created a product based on it. It it had it it it it used it benefited from Westlaw's value added. And so the judge was was very direct and said, none of Ross's possible defenses hold water. I reject them all.

Wow. So that's that's quite a turnaround from, last year concluding very differently. So that's that's interesting to me. So I haven't read anybody yet about what the implications of this are for the many, many other cases that are out there now. Right.

Yeah. 38 at least 38 AI related copyright cases making their way through the courts right now. I'm I'm amused every time now, of course, I read the New York Times and they mentioned, OpenAI. They have to say that the New York Times is suing OpenAI in every damn case. And OpenAI is in the news constantly, and it amuses me.

But Yeah. So in this case so then is it a matter of taking that information and regurgitating that information as if as if it was theirs and that would be the difference versus training a data training on a dataset that contains that information, but the AI or the the LLM or whatever you're using is interpreting that to mean things that it devices. You know what I mean? Like like, in this case, was it truly, like, we use that dataset and we kind of regurgitated it, and that's the problem? Yeah.

I I think so. And do not. I'm no lawyer. Right. And I don't I don't know I don't know Jack, but I I think that's my my guesswork here is that, if you're using text broadly speaking to train a model and you're literally training it to speak, you're training it to know this word goes after that word no matter what the context is.

In some cases, it's more context specific because certain words like litigation only occur generally in certain contexts. Yeah. But all in all, you're teaching it to talk and listen. That, I I think, at a high level still strikes me as fair use. In this case, it was a very specific application where this was a, something you could come in and use this instead of Westlaw.

That's the other issue with fair use. If it is it has to be transformative and it can't be, it cannot replace. If it's a one for one replacement, that's obviously an issue. Yeah. So, interesting.

Yeah. Alright. Well, I mean, notable for, for the fact that it's kind of one of the earlier copyright cases related to artificial intelligence, notable for the fact that, you know, if that being the case, it is an example of, you know, fair use not working. But it sounds like to me, from what you're saying and from what I've read through here and everything, that you, yeah, that it's probably not necessarily a bellwether for some of the future cases, that we've talked about a lot on this show that, you know, have to do a little bit more with using the information to train, on the back end and, and making that transformative. This is, this seems like it's a little bit different, but still notable.

Yeah. And a lot of it, this is about the, the head notes and key number system in Westlaw. That's quite specific to this case. Oh, very. Yeah.

And so I'll put it I'll put something in the rundown you could add into, our our our own show notes, from a a law firm that's writing about this I just found. So Oh, okay. Cool. Excellent. Yeah.

You can just go ahead and drop that in under there. And I will definitely include that in our show notes, which you can find at AIinside.show. Often just attached to the podcast if you download it or watch it on YouTube. We put we put them everywhere. We throw them all over the place.

Let's see here. Okay. If you thought DeepSeek was shifting the kind of approach of big spenders and big tech on artificial intelligence and, kind of driving the cost down as we've been talking about, and people have been surmising that, oh, well, DeepSeek proves you don't need to spend a lot of money. How is that going to directly impact these big tech companies that are throwing massive amounts of weight and and money behind their, you know, the training of their models and their domination in this race that is very much on. Well, Microsoft, Google, Meta, all projected to increase their spending on AI by 45% from last year.

It's a 215,000,000,000 combined total at least, according to the Wall Street Journal article that I have right here that I can show. Amazon also expected to increase their spend largely due to AI. Andy Jassy, who is, Amazon's, CEO, said in an earnings call last Thursday, quote, we think virtually every application that we know of today is going to be reinvented with AI inside of it. There it was. There's the title of the podcast.

Hey. AI inside. I didn't even mean for that to be Let's sue them for trademark violation. Yeah. You go down Amazon.

We're gonna make at least $10 from you. And Sundar Pichai, Google's CEO said AI is quote as big as it comes, and that's why you're seeing us invest to meet that moment. I mean, all these big companies just realizing, you know, the the, the war is on. They they all do not wanna be left behind. They all want to be at the head of the pack.

And if DeepSeek was applying you know, was thought to possibly apply pressure on them to not feel like they need to spend a lot of money, think again. It doesn't seem like that's the case. But they probably planned all of this well before the DeepSeek thing happened. And and they're and we're not sure how cheap DeepSeq really was. We've talked about that on the show.

Well, that's true. There's doubts about that now. Right? Exactly. And so, so so when I when I worked at Time Inc and I started the magazine Entertainment Weekly, I provided myself on starting it, very penuriously.

I I had less of a staff. This believe it or not, this is gonna get relevant in a second. Just give me a second. I had a smaller staff for a weekly magazine than any company in any magazine in the company for a monthly magazine. I I saved, $3,000,000 a year on technology because we were the first one to use Macintosh's to produce everything with my wife who set that all up.

I prided myself on saving all kinds of money. And then some wise old sage said, Jarvis, nobody ever got fired here for spending too much money. And I think that that's the, that's the attitude in Silicon Valley is they don't wanna look back. And some analysts saying, well, if they if they hadn't pulled back, if they had if only spent more, if they'd only invested more, if they only had the guts, if they only had the courage, they'd be better off right now. They don't want that in this, gonad war, where we are right now.

And it's not just these companies. It's also, Macron in France saying we're gonna spend a hundred billion dollars, whatever the number was. And and and that goes as well that, everybody's trying to spend money now on this to prove that they can be the best and the biggest. You know, we'll see. We'll see what happens.

Absolutely. And then, that's so funny. Like, no one ever lost a job by by spending more money. That's that's so true. Yeah.

That's that's so interesting. Seems seems, like, perfect with where they're at right now. But I found I also came across another article, Wall Street Journal article, that was kinda pointing out that the investment that we're seeing is kinda counterpointed with the fact that IT unemployment is rising. Meta let go of 4,000 people. That was a a bit of news that was rumored to happen, at the beginning of this week.

You know, over the weekend, it was like, a minute is gonna lay off. And and sure enough, you know, they they laid off, I think, 4,000 people, 5% of the workforce. And I think that's more telling than a projection of spending because you can always cut back on the spending. You can always choose not to sign something. It's it's just something you're telling the market to get the market, pre prepared for the cost.

But when you lay off 4,000 people from one area and shift those resources to another area, that's a more telling story, I think, about where the strategy is. And so Meta is trying to clear desks for AI people. And, and I think that says that's where their priority is. Right. Yeah.

It really seems that way. Goo Google, of course, we we've talked about offering buyouts for its employees, to prioritize AI. Salesforce, we hadn't talked about reportedly laying off a thousand people Mhmm. And making way once again, you know, for hiring salespeople for its AI products. And evidently, I heard I saw some story that, that, Benny office said that every sales call has to include AI agents now.

And it was only a year and a half ago, I heard a a a Salesforce executive in San Francisco say, we're being cautious about agents because we're not sure we trust them yet. Well, that sure changed overnight. Apparently, we trust them a lot now. Everywhere. Everywhere.

They need to be everywhere. And then I I mentioned the Wall Street Journal article, and then I didn't really talk about it much. But, IT, unemployment is rising 5.7%. They say actually rose from 3.9% in December to 5.7% in January, passing the overall unemployment rate in the process. And, you know, largely because AI is impacting the labor market in IT as well.

So it's gonna, it's gonna tackle that area. It's the first on the mountain for the avalanche. Yeah. Yeah. Interesting stuff.

Not not that it's surprising that AI would would have those impacts, but, you know, I I guess the question that I have out of something like that, especially when you're talking about people who are technologically minded like IT workers, is, you know, does that does that ultimately long term put them out of a job, change careers, or does that, you know, or is that kind of a nudge for them to transform their their job and their skill set to work with AI, to to be skilled in that regard so that they can continue doing what they were doing, but now through the new kind of lens of what that means? Yeah. It's a it's a it's a really good question, Jason, because I think, even there to think that, well, I've gotta remake myself now into AI land Mhmm. May be a bit of a mistake. So I had a I had a wonderful, conversation last Saturday last Friday with, Mike and Marcy McHugh, the founders of Flipboard.

Okay. And they have created surf, which this is not directly an AI topic, but I'll I'll get there in a second. I keep on doing the show. Well, I may seem irrelevant, but I'm gonna try to find relevance. No.

No. No. So surf is a really great, combination of the, open protocols in social. So from the one place it's it's in private beta now. One place, I could interact with people people on Blue Sky and on Mastodon, and I can read ghost newsletters, and I can watch videos, and I can do all these things from that one place because they are open these are open protocols.

And my point finally is that I think because of the fall of Twitter and I believe, Facebook and Instagram, we see a lot of action in social. And I even think, I was talking to an executive at Mastodon also last week, and I said that part of Mastodon's appeal might be that unlike Reddit, you know, Eugene can't just sell everything on Mastodon to OpenAI. Sure. Because it's it's federated, and he doesn't control it. And that's the advantage.

And it's kind of your space that's free for me. I know anybody can scrape it, but nobody's gonna make money on your stuff. My point is that social still lives. Social used to be the whole web not long ago. Before that, search was the whole web.

But we still have search, and we still have social. And these things are additive. And so I think that people who lost their jobs because they're not AI, it may be very smart to go and learn more about AI. It may be smarter to see what's affected by it or what's being, neglected these days. And I don't know if I were advising a former student, on such a thing.

I'm not sure what I would think. Yeah. You know, it can go either way. And what does it mean to learn AI now when it's when at one point at one level, it's very high level math in the making the foundation models. But otherwise, it's anybody's a programmer now.

Anybody can do it now. So I'm not I'm not sure where this where this really goes. Yeah. What does that's a really good question. What does it actually mean to learn AI?

Let me ask you this, Jason. If you're knowing what this is doing to computer programming, I don't know what your daughter's interests in life are. But if one of them were, going to college Yeah. And asked you not that at that age, they ever asked us parents. If they asked you, whether they should major in computer science, what would you say?

Oh, that's such a good question, Jeff. I mean, I don't I don't necessarily operate from the assumption that computer science is is, you know, some like, some bad thing to put your time into. I think that there's plenty of of of distance to be gone with computer science. And I I just think what it what the target or what the focus of that education could be is different now than it was even five years ago. As long as you're as long as you're moving with the, you know, with the general kind of wave of everything, I don't see a problem with it.

And I'm and, also, just like at at my core, I'm just kind of the kind of person to be like, well, if you're passionate about that, then you need to follow it. So that's Yeah. But but is that a wise decision? Is it wise for them to study it? I think it still is.

I don't see why not. Yeah. I I I think you're generally right. But but, you know, when we had our our preparatory shows, we talked to an executive who said, oh, no. It's it's it was not.

It was Jensen Wong who said, that universities should should stop training computer scientists. Mhmm. I think everybody would say that's going too far. But how many jobs is there gonna be? You know, it was like there was not long ago where if you if you got a law degree, you were safe.

If you got an MBA, you were safe. That's not the case anymore. If you have a CS degree, you're not necessarily your employment is not necessarily insured anymore. Let's just say that. No.

No. But, you know, Chris, who's watching live stream says, computer science has AI courses. Oh, absolutely. Absolutely. Absolutely.

The AI tracks, Kevin Koeppler. Thank you everybody for so many sudden comments. I'm focusing on AI data center infrastructure seems to grow. That's gonna grow. You bet, Kevin.

Yeah. And, yeah. And I think that's that kinda gets to my my thought about that, which is as a blanket recommendation, I wouldn't feel comfortable saying don't do that. No. No.

Because I think there's a there's a way to find it. There's a way to find a connection that actually still continues to be relevant. I don't know. Chris says it's right, is that it's about theory and math. And I think that's, that's a really important structure.

My son, who is a developer at a young age, he sold a, an application to Facebook when he was 14. And I worry about he was gonna do that. He's not doing that now. But, he's helping kids now. But, when he started in computer science in college, he was kinda disappointed because it was so much math.

And what he wanted to do was build things. Yes. And many CS programs are not building things. They're about math. Now the math is always gonna be useful, I think.

Mhmm. For sure. For sure. Your son sold at what? At what age?

He sold an application? At 14. There was a when for a while, Facebook thought it was gonna be a platform and they opened up for applications and they even killed some of their applications so that you could do it. I forgot what it was exactly. It was a it was a it was a a way to to to, I think, share your classes.

That's cool. Yeah. And he replaced it and did it, and it got and it got ordered because it didn't get sold to Facebook. I'm sorry. It got sold to a company.

Uh-huh. Yeah. That's pretty cool. It was. Wow.

What an achievement for a 14 year old. Yeah. That's amazing. That is super cool. Cool.

Excellent discussion there. A bit of a Google roundup I thought I'd throw in real quick. First of all, this is only slightly and and quickly, AI related, but Google IO announced it's happening May. Four key areas, artificial intelligence, Android, search, Chrome, And, you know, there's hints, on the site, the IO twenty twenty five site that there's going to be, you know, stuff focusing on Gemma, the Gemma open model, Google AI Studio, NotebookLM, that sort of stuff. Also, this is interesting.

It happens during the same week as Microsoft Build where you know Microsoft's gonna be talking about its own AI efforts. So that's gonna be a big news week. There's gonna be a lot to lot to discuss. Are you planning to go to one of them? I am absolutely going to Google IO.

I already got my invite. I already accepted. Oh. And so I will there. That's good.

For sure. Wouldn't wouldn't miss it. When they when they give me the invite, I'm I'm an immediate yes. So Yeah. Used to be we got nice goodies like a phone.

That would that that that doesn't happen in recent years. You never know. You just You never know. You never know. Oh, the hope's always always there, Jason?

Now I did mention notebook l m. And speaking of, there's some, real quick news there. Notebook l m plus now available with Google One AI premium for $20 a month. So the notebook l m product that's free that everybody can use, that doesn't change at all. Google's really focused on just adding more capabilities into the plus, so the baseline will stay free according to Google.

Like, you could make, like, unlimited podcasts and stuff. Yeah. Right. Like, you know, if you're plus, then you get five times as many audio overviews, 20 per day versus three. Do you really need more than 20 audio overviews per day?

Right. Up to 500 notebooks versus 100. You can add 300 sources to a notebook versus 50. I mean, but the baseline that they give you for free is pretty solid. Pretty great.

It's pretty great. Yeah. The the the the the main thing that I find is, when I've got into the to the original, I think it was in beta. I could only add, like, two documents, and that was very limiting. To be able to have 300 documents in for researchers, which should be able to then go and query all of that.

Which one said this? Who disagreed with that? That's valuable. That's very valuable. That's a part time job collecting 300 sources.

Well, so understanding it all. Let's see here. This pile of paper Uh-huh. Is for two paragraphs. Oh my goodness.

That's, wow. I would not have guessed that. I've always had such respect for people who who can get their thoughts together and organized for a while. For the the previous section of about 10 paragraphs. So I'm now writing about the line of type.

That's my next book. And I'm in the chapter about its impact on on, labor. And you it could pile up real fast. I just because I go down another rabbit hole and I find, you find a citation. This is how real academics do it.

I'm not a real academic. Oh, there's a citation I gotta follow. There's another one I gotta follow. There's another one I gotta follow. And then because I wanna mark them up, I print out like crazy.

I'm surrounded by these papers and, other other documents. So if I if I had started this if if notebook LM were there when I started this book. Yeah. Well, I might have done it differently. Well and, I mean, it does exist now.

So maybe you get on the other side of this book and whatever your next book is, whenever that may be. That's what I'm thinking. That'll be an interesting because I have a next one in mind, and there'll be a lot of, papers and theory and things that I'm gonna, wanna put in there to try to get some sense of. And so, yeah, I probably will do a lot more, making sure I have PDFs of them and rather than printing them out, maybe marking them up in on the screen and making sure they're all in notebook l m. And I organize them that way.

Mainly not not because I wanted to write anything, but I wanna be able to query it. You wanna be able to yeah. Who said this? Who disagrees with that? Right.

No. That stuff is really I remember this quote. Which where where the hell was it? That's a simple search, but it's but it's easier. Who who argues this point?

Mhmm. Right. That would be possible there. So I, but it means that I've got to use my, my beloved scanner. So certain books I will end up scanning in just so I can get them into a book l m.

So so any of the time that you save in one aspect is like is removed on the other side because, you know, you've gotta plug the analog hole, I guess. Yeah. Yeah. That's so interesting. I'll be curious to hear.

I'm sure you'll talk about it, but I'll be curious to hear kinda how that goes, you know, if you find at the end of the day that it was useful or if it was just kind of shifting your work from one part to another. You know? That's been my experience with AI is it makes certain things really helpful, but it doesn't necessarily save me a ton of time because I've said it in other ways. Yep. Yep.

Okay. That's related. Yeah. And then finally, a new AI mode, potentially coming to search. Thought this was interesting because we've already got like like, we were just talking about notebook l m, which is really great for research.

We've got all these these tools that are starting to kind of, you know, make themselves known. We've got deep research. You know, I've talked about perplexity in the past. All these, you know, research, oriented AI tools are, you know, certainly a thing right now. And so you would think or I would think that notebook LM is kind of what that you know, what Google's effort would be around that, but apparently that's different.

This AI mode coming to search potentially, it's being dogfooded at Google right now, and it, quote, search intelligently researching for you. So organizing information into easy to digest breakdowns with links to explore content across the web. It's kinda like a Google search research tool. So you might be able to ask, and this is a query, you know, that Google gave as an example, how many boxes of spaghetti should I buy to feed six adults and 10 children and have enough for seconds? Or, you know, compare wool down and synthetic jackets in terms of insulation, water resistance, and durability.

And I'm sure you could ask it to put it into a a table or something like that. We're just seeing more and more of this stuff, and this seems to be a version in potentially integrated into Google, or I don't know if it's a certain mode that you snap into that's separate from search, or whatever it is. But, yeah. But interesting thing here, a few fold one. So, so the, the, the, the, the, what am I trying to say?

Taxonomy? No. The, the, the, the history, I think of media It is where I put it as we spent a century in, in, in, destination mode, come to us by our publication. Then we had search and that generated SEO. Then we had social and that generated, audience development.

Now we have queries generative and that's GEO generative engine optimization, and then agents we think. And so there's a few things. One is how do media properties? I just had a meeting with a big media company last week where there's, how do we deal with this? You know, and I said, your issue is discoverability.

How do, how do you know that this machine is going to call upon your material? How do you make it so they can call upon it and how do you get branded for it? What benefit do you get out of that? That's big questions. Right?

But it's also true. We talked to Shibsted in our second show. They were doing that exact same thing on top of their content. And so you can query, that was just a test, but you could query all of Shibsted product reviews, and get into a conversation with it. So I think we're going to find that you're going to see a micro and a macro version of these searches that within a brand, that's a new way to get to information rather than just reading stories.

And then the macro version is across many brands, Google, you replicate, that. So where this comes out in the end, I have no idea. Yeah. Yeah. But I mean but it there one thing is for sure.

And Pichai hinted at this, I think, earlier this week said, 2025 is, quote, one of the biggest years for search innovation yet. Dun dun dun. Is he talking about something like this? I mean, the search product as we've known it from Google is really is changing, and, that's there's no doubt about that. Whether it's changing for the better or for the worse, that's the Well, and the question we always ask is hallucination so called and randomness.

Will you get the same answer twice in a row? No, you won't. Does that affect the credibility of the search? We'll see. Right.

Yeah. We absolutely will. Alright. Quick break. And then when we come back, got a few stories to round things out, including a, an article in the MIT Technology Review, that is talking all about AI crawlers and what it does to the web.

That's coming up here in a second. All right. I thought this article was, interesting. You put this in here, so thank you for doing that. And I think we've talked a little bit about these topics in the past, but MIT Technology Review, has an article analyzing the conflict brewing between websites and AI companies, specifically as it relates to AI crawlers and kind of the, I don't know, kind of like the the arms race or the the war that is the back and forth and what that does to potentially deteriorate how the how the web operates over time.

Yeah. The the issue is, as I understand it, back in the early days of search and search engine optimization, your aim was to be crawled. You would do anything to have Google crawl you, right? Cause you thought the only way you were discovered. The only way you could be found.

You, you wanted crawling desperately. Now because of AI crawlers have cooties. And so if you start cutting off the crawler just for AI, but then you start to say crawlers are bad, then you're not only hurting your own discoverability on the web, you're also hurting search results and thus the primary entry into the web. Right. And, and this, this was something that where, I had an event as you know, about more than a year ago with the common crawl foundation, which has an open open source crawl of the web, cited in more than 10,000, publications.

We had a it was our first show we had on the the head of, Common Crawl, I think. That was our first show. Episode one. Yep. Yep.

And, so you can go back and and and and and and watch that if you'd like. And so crawling was a good thing. It was the way we found things on the Internet. And if crawlers have cooties and people start cutting them off wholesale, then, that affects all of us. And we lose the open ethos of the web, and that worries me greatly.

So we'll see where this goes. I say at the end of every story, I say, we'll see. We'll see where this goes. But, but I think it was a very good catch on MIT technology review to look at that higher level question about what's happening with how the web now operates. Yeah.

Right. Can the open web survive? Who controls the information is another thing that they point out kind of the, the concentration of of control over the information of the big the biggies in the room versus the small players, small content creators, you know, hiding their content behind logins or taking it completely offline because they're concerned about what happens to it. And, yeah, that's an interesting thing to consider. And then, you know, always also coming back to another topic that we talk about frequently, which is the quality.

Like, we do want quality artificial intelligence systems that we can use. I mean, if you're using them, they have so much potential. But if they're not being fed high quality information, then they lose a lot of of that potential. And, that's not what we want either. So I don't know what the right right answer is there.

And then speaking of agents that crawl around the Internet, Jeffrey Fowler from the Washington Post, who I've always really I've always really liked his writing and kind of his approach on things. He played around with ChatGPT's operator in his article. He, asked it to do a few things. This operator is ChatGPT's agent, essentially, browser agent. He asked it to find cheap eggs nearby, and the agent went out, it researched, and then used his credit card to buy a dozen eggs for $31.43 and have them delivered to his house.

And he said, I'm overpaid by a dollar these days. Yeah. I mean, eggs are expensive. I don't know if it doesn't eggs at $31. No.

Almost $32 at this point. On the flip side, Jeffrey also asked operator to find him a less expensive Comcast plan, which it did identify. And it also read through the five fine prints to note that, you know, eventually on this date, the price would would go up, and that enabled Jeffrey to decide whether that was actually a good deal. And I think he ultimately passed on it anyways. Yeah.

I think and I think what really kind of jumped out to me here is kind of the idea that it's, agents do things for us. In a perfect world, an AI agent acts in our on our behalf in the ways that we would want it to. So it doesn't, you know, do things that we wouldn't want for it to do on our behalf. Right? But in order to do that, it really needs to know so much about you.

And that's one aspect, not just behavior, not just preferences, which are all really useful and beneficial for an agent to truly operate in your, you know, on your behalf, but also things like authentication and passwords. And so truly everything and You have to trust it. You have to trust it with all that information. You have to trust that that information is only for this agent and doesn't go back to the mothership or whatever. And, you know, whether the architecture is even set up to do that to begin with or not, that's gonna be a real big hurdle for a lot of people is just because they won't understand why that is or is not the case.

And it really kind of speaks to this idea that seems to be a recurring theme with technology that I've noticed, which is that we w we say we want all these things. And then when it comes down to it, we realize that in order to have all these things, we have to give up those things. And we say we don't want to give up those things. And so I don't know where it leads. We'll see, as I like to say today.

Yeah. Yeah. I think so. And, and clearly with agents, you're going to have to check its moves as it goes. And so there's this whole notion of whether or not we're actually seeing reasoning from these models.

Certainly with agents, you're going to see steps. And just to take, you know, to, to get the hens off the hook here, for his eggs, operator found a dozen dozen large white eggs, not even organic for $13.19. Now they're as high as $7 in some places now, but that's almost double. So that's bad. Right?

But then, it purchased, it purchased those adding a $3 tip, a $3 priority fee on top of that, on top of the $7.99 delivery fee on Instacart and a $4 services fee. I'm not sure what that is. And a 25¢ bag fee. Oh my. So at some point in that process, it could have come come to you and said, Jeff, you sure?

Or if it were really smart, it would go and say, I'm finding the average price of eggs now is this. Do you you need them that badly? I'm I'm finding that the average eggs, eggs price is this. I've noticed in your calendar that you have twenty minutes open, and your nearest grocery store has them in the store for $5. Why don't you, in that twenty minute time, drive down there and pick up the eggs?

Right. Right. Great. Thank you, agent. You just saved me $30 or whatever.

So, anyway, it's fun. I'm I'm glad he's playing with it and and Yeah. Yeah. Yeah. And, so that's that's fun.

A lot a lot of the articles that Jeffrey, writes tend to be a little kinda on the playful side Yeah. Of this sort of stuff. Way to put it. I always appreciate that. So so a good day.

Good rundown. Lots of good stuff. Times. Yes. And, thank you, Jeff, for for being here and for talking about all this fun stuff with me each and every week.

And we talked about your your authoring, of of a book that has not been written yet or is in progress, but we have a couple that are already written, the Web We Weave, Gutenberg Parenthesis Magazine, all at jeffjarvis.com. People should definitely go there and check it out. And I'm super curious to hear how your process goes as you're Yeah. Thanks. As you're integrating.

Yes. Super interested. I had mentioned not too long ago that, we have a new YouTube channel for AI Inside podcast, and we do AI Inside show. So if you go to YouTube, you can see it's a it's a channel dedicated just to this show, so that's pretty awesome. We've got it all set up.

Oh, I love how, like, because we're live, it's got a nice red Yeah. Isn't that nice? Live thing around. I I had never noticed that before. So it looks really nice.

Go there. Subscribe. We're kind of building up the subscriber base for the YouTube channel. That's important for what we're doing. So thank you for that.

You'll get notified when we go live like we did today. Very last minute. I forgot to create the event in advance. Also aiinside.show is the webpage that collects everything that we do. You know, it's got links to all the things it's got, it does have the audio podcast, but it also has, you know, embedded video player for the video version.

So if you really just need one place to go to find everything that we're doing, you can go there. And then finally, patreon.com/aiinside show. If you wanna support us on a deeper level, you can certainly do that. And we have some amazing people supporting us on that level. DrDew, Jeffrey Marraccini, WPVM 103.7 in Asheville, North Carolina, Dante Saint James, and our newest executive producer, Bono De Rick. Thank you so much. Thank you. Bono, good to have you all on board. Y'all are gonna get a if you haven't already, get a get a t shirt and the, access to the Discord and ad free shows and all the things that we do, and also the pride knowing that you're helping us.

And our gratitude. Thank you. And our deep, deep gratitude. So thank you everybody for watching and listening. We love doing the show, and we will see you next time on another episode of AI Inside.

Bye, everybody.