China’s Practical AI, AI Crawler Clicks, AI Music & Jobs Impact
Hi, everyone, and welcome to the monkey patching podcast where we go bananas about all things, China influencers and more. My name is Marillo, joined by my great friend Bart.
Bart:Hey, Bart. Hey, Marillo.
Murilo:How are you doing? I am doing good. Doing good. And you're older, as you know. So still adjusting to the old age, but doing good.
Bart:Yeah. Another you're 30 now.
Murilo:Mhmm. How does it feel? New decade. Yeah. It feels feels good.
Murilo:I feel like I feel good, but I feel like I should be having a midlife crisis or third life crisis or whatever. But I'm not. I'm still strong, ignoring the haters. You know?
Bart:It'll come. It'll come.
Murilo:It'll come. You're speaking from experience. I'm like, don't rush. You'll get there. You will get there.
Murilo:How are you doing, Bart?
Bart:I'm doing good. I'm doing good. Okay. Little bit of a hectic week, but that's okay. It's good to be here to have a bit of a hectic week every now and then.
Bart:Also, like, also a hectic week in news, but that maybe that's every time. Right? Every every episode we have, it's been a hectic week in It's true.
Murilo:True. It's true. AI, like, it's the world goes the the world going bananas about these things. Pun intended. So what is the first thing?
Murilo:Should you start or should I start?
Bart:Oh, give it a go. Google rolled out a DeepMind built image editing model inside the Gemini app aimed at keeping people and pets looking consistently like themselves. It supports blending photos, multi turn tweaks, and playful scenarios like, for example, putting a tutu on your Chihuahua. For everyday users, that means more believable outfit swaps and room makeovers, plus visible and send ID watermarks to clearly flag AI generated edits. It's it's called a Gemini 2.5 flash image, and their code name the code name is, I think, Nano Banana, if I'm
Murilo:not sure. I was looking. Like, in the top on the the browser says Nano Banana. Image it being Nano Banana.
Bart:And, actually, it's apparently, it was already on a la marina under the code name Nano Banana. Nobody knew what it was where it was from, but, apparently, it was Gemini 2.5. And now I saw that in the Gemini app, they renamed it again to nana banana. So apparently, the there there was so much hype around nana banana that it became important to actually also name it like that in the Gemini app.
Murilo:Yeah. Everyone was like, we want banana back. Maybe, like, maybe it's all the the monkey patching podcast fans that, like, rallied around Yeah. To make this change. True.
Murilo:You know?
Bart:And and my first time trying it was on a picture of you a bit in line with the code name nano banana. I I I asked this, like, here is this guy uploaded a beautiful picture of you, and I I I asked Gemini to put a banana in your mouth. And it it looked like someone took a picture with you with a banana halfway up your throat.
Murilo:I'm resisting my my urge to derail the conversation and make comments.
Bart:Yeah. Let's not do that. Let's not do that. That's not I'm just talking about photo editing. But it's really like we're we're laughing a bit, but it's to me, like, it's it is a major step up from what the state of the art was.
Bart:To me, this is a this is a major breakthrough, and it's mainly, like, because it's very precise in what you want to edit. Like, normally, if I would have a picture of you
Murilo:But and then editing is just by prompting or do you select
Bart:By prompting. Yes.
Murilo:Part of the image? Ah, okay.
Bart:Just by prompting. Yeah. And, like, most of these these type of models before this was, like, I would upload upload a picture of you. I would I would ask, put a banana in this guy's mouth. Then I would probably get a picture back of a guy with a banana in his mouth.
Murilo:Or maybe not even on his mouth. Maybe, like, on top of
Bart:his face. Guy would maybe have some of your facial features,
Murilo:but it would probably not
Bart:be you. Right? Like, the it was it was very hard to, like, keep consistent in what is already already in an image and only change what you want to change. And this works for people. Like, I've tried it also later with, like, changing backgrounds behind people and stuff like this.
Bart:Works very well. But also, like, you can also say, like, this part of the this this this color of that specific furniture in this room, change it to that color or even change it to a different material. Wow. And it works really remarkably well.
Murilo:I remember that yeah. And my current employer, your former employer, let's say. I don't know if employer is the right word, but anyways. We also did projects with, like, styling parts of the furniture. And back then, it was, like, really I was like, wow.
Murilo:It was wow, but, like, the results were very different. It was like
Bart:had look at the colors, like, the very disappointing. Right?
Murilo:But at that point, it also wow. Yeah. At that point, it was like
Bart:chair and this really, and the color that I wanted.
Murilo:You can kinda see oh, wow. Like, it was really, like, too you had to squeeze, try it and imagine. But now it's, like, now it's really like, now it looks like the picture was like that. Right?
Bart:Yeah. Now it's there. And to me, like, if it like, I'm saying, like, to me, this is, a major step in the evolution of these models. Like, this is such a big difference from what we had a few weeks ago. And that also in a context where everybody for the last year has been saying that everyone and there that there are no major breakthroughs anymore.
Bart:Right?
Murilo:Yeah. Yeah.
Bart:Yeah. Yeah. Is only very, very small changes. But to me, this is a major breakthrough, and it's the same as what we discussed the last episode about the three d world models.
Murilo:Yeah. That's true.
Bart:Can still see that there are these major breakthroughs that are that have a big or can have a very big impact.
Murilo:I think the so and maybe to for the people listening only, we're talking exclusively about realistic images. Right? It's not we're not talking about cartoons. We're not talking about animations.
Bart:It's Right. With I was talking about realistic image images.
Murilo:But I think that's the thing that was and I I mean, I didn't try. I mean, I didn't see any examples. I didn't try either, but I think today, there's a lot of good models that do cartoon or anime style or these things. And I think the realistic images, that's when I I also felt they fell a bit short, and I can this always the from your experience for the picture I saw the pictures I the examples I saw as well. This one looks very impressive.
Murilo:Right?
Bart:And everybody with a with a Google account can try it for free if you go to the to the Gemini app, either on your phone or in the browser. They're saying it's the usage is kept, but I've there have been a few sessions where I've used it extensively. I haven't reached the cap yet, and I'm not on a paid Gemini account at this moment. Probably, like, they're very lenient at this point just to get a lot of user interaction. Yeah.
Bart:I can imagine that this that's worth something to them as well. But definitely try it out.
Murilo:Indeed. It looks it looks very cool. And I think also in in light what you're saying about the the major breakthroughs, it also is looking from Google. Right? The Genie three, I think, was the first role simulation, and the the breakthroughs look like they came from, not the text, right, from these other domains.
Murilo:When I think and I think they also in this article here that I just skimmed through, they also give some examples of how you could possibly use these things. Right? Like, if you have a, I don't know, for fashion. Right? You wanna say, put this person with this dress, now put them with this dress, and then, like, just kinda see how you will look like with this or that.
Murilo:So I feel like you could even get like, this could even move forward the the application space for Yep. Image generated with AI.
Bart:Think it will also fuck up some of the startup that have very much focused on this. Like, like, outfit swaps, like, optimizing portrait images, like these kind of things that you now just get out of the box with this.
Murilo:It's true. And I think it's the when you say that, it reminds me a lot of early days of Chai GPT, how every time Che Gpt released a new feature, it would kill, like, hundreds of startups. That's right. Mhmm. Yeah.
Murilo:So I agree. I agree. What else do we have? We have China is running a different AI race, prioritizing practical low cost systems over America's moonshot chase for AGI, so artificial general intelligence. Beijing is steering funding towards startups, more smaller data centers, and rapid rollouts across services from agriculture to health care.
Murilo:If superintelligence remains far off, this today first approach could win, putting more everyday AI in our products long before any big breakthrough. So I had a look on the the article as well from what I what I gathered. Right? Like, basically, the differences, they're they're they're emphasizing how The US is really chasing the AGI. They're saying there's a lot of military value, all these things.
Bart:This moonshot goal, like, let's put everything out there. This is what we need to achieve. Yeah. Everything is is is, like, in function of achieving AGI.
Murilo:Exactly. From what the article goes on is the that China has more on the implementation part. Like, it's not about having the AGI model, but it's about making sure that the stuff we have is already embedded in a lot of places like hospital and agriculture. And it also highlights a bit how in China, they government is also sponsoring. They have a lot of money being invested on these startups to increase the adoption or, like, the embedding of of of the AI that we have today.
Murilo:When I read this through as well, I was also wondering how do we define AGI? It's a bit tricky. Because, I mean, I know they say, like, AGI could help with, like, military breakthroughs and, like, cure cancer and all these things. But to me, like, hey. How do you know when you get to AGI, how do we know we got there?
Bart:Yeah. It's a difficult definition.
Murilo:Because, like like, before we had the the the I think it was the Turing test. Right? Which, like, the that was the first time they were thinking of AGI. Right? Like, the and the idea was if you the test was and correct me if I'm wrong here.
Murilo:Like, you have someone talking to a chatbot basically, and then you have a computer and you have a human. And if the person cannot tell the difference between the human and the computer, that's AGI or not AGI. They didn't call AGI then, but that was, like, true artificial intelligence. Yeah. Yeah.
Murilo:Where are we past that?
Bart:I would expect this, brother.
Murilo:Right? So it's like now AGI is saying, like, okay. AGI is when you have an AI that discovers the cure for cancer. It's a bit
Bart:that can work independently and alone and can find a purpose. Yes. And and the thing here, like, with article highlights is basically, like, US, everybody's looks to be a function of reaching this AGI while China today focuses more on what can we do with the AI that we have today and really putting that much more in practice. And I think that, can we all also see, like, this AGI timeline, like, keeps slipping. We hear a statement every time that it's gonna be next GBT release will be AGI or at least hints towards that.
Bart:Right? That is never there. Yeah. And, like, this China is is taking more or or that's at least a bit of the messaging that you get is like this. It's good enough when we try to put it everywhere approach.
Bart:I think that is it's interesting we want to see, and, apparently, there's also, like, it's a little bit of a a reaction as well to, the chip export limits. There's also there there's some some messaging on that. It makes it for them also just simply harder to superscale the training. So let's focus on what we can do with it today. So there's under and I I think why people are talking about this now is also because we had the, what's it called, world humanoid humanoid robot games two weeks ago, want to say, a week ago in China.
Bart:It was held in Beijing where there were, like, I think something like 500 humanized robots. And it really shows, like, what is already possible when you combine AI, for example, with robots.
Murilo:And it's
Bart:it was super impressive. And I think that's also what made people talk. I mean, I think everybody was a bit surprised, like, of the performance of these robot and these humanoid robot games. And, like, also because of that, like, you also see this China is able to put AI in things like robots, but it's also, like, it has this massive supply side push. So, apparently, one of the robots that was that was participating in the event, I think it's it's from unit three at the company.
Bart:You can buy it for €6,000 Ah. Which was yeah. And if you look at look at so you on YouTube, what that can do, it's it's unimaginable, but it's it's it was unimaginable a few years ago. And I honestly, I thought this was this would be unimaginable today to hit us for €6,000. But apparently, China is able to do it, and I think, it is potentially because of this different type of approach.
Bart:Like, let's make sure that we can do something with it today. And one of the things that it apparently is also playing because it's, like, a bit like China's a bit like, quote, unquote, state sponsored capitalism. And the challenge that they have very much in China, which we also have in Europe, is that there is, I think, a lack of VC funding when you compare this to the states. But the way that China's solving that is that, basically, Beijing is filling this gap with public money, and it looks like they're they're achieving this. Right?
Bart:Like Yeah. I They're successful in this.
Murilo:I think they also in the article, they also bring other examples of, like, high school and also hospitals and also agriculture. Right? So they're really embedding across a lot of places. And to me, this also reminded me a bit
Bart:Maybe the hospitals, but I haven't actually looked at it. I heard it in an interview. There is apparently a full AI driven robot hospital in Beijing where you can go, and there is, like, no person there, like, all robots.
Murilo:Yeah. I saw that. I I mean, I saw I I think I I read that they were building something like this, but, yeah, indeed, it was, like, fully fully fully Yeah. Robotic. Yeah.
Murilo:Yeah. Yeah. This also makes me think of the the difference between, like, the the research and the engineering or, like, the, you know, like, how one is really pushing the boundaries to see what's possible, and the other one is pushing the boundaries to see what how we can make this useful. Right?
Bart:Yeah. Well, there's also mean, I think it's maybe the opposite statement that you're making here is, like, there's also, like, there's there's some, well, there's a lot of theorizing why how come the China is very so very good at this. There's also, like, a lot of key players, and I think it's around, if I'm not mistaken, around 30% of, like, the key, like like, figures of the state in China are engineers by background.
Murilo:Didn't know that.
Bart:There are it's way, way, way less in The US. And I think if we look at, for example, Belgium today, it's zero. And this might also play
Murilo:in in Yeah.
Bart:Into these sort of things. Yeah.
Murilo:Probably does. Right? I think it probably does. I think also even just the fact that you understand some, like, of these things when you're discussing it. Right?
Murilo:It makes I mean, how would you imagine it makes a huge difference? And the the title of this article again is, like, China has a different vision for AI and might be smarter. Do you think it is smarter? What is your opinion on this?
Bart:I I it's a difficult one. But to me, the difficult thing is also, like, it's one doesn't go without the other.
Murilo:Yeah. That's true.
Bart:I think, for example, we wouldn't have things like DeepSeek and what was the other one? That would key Kiwi or something?
Murilo:Kiwi. There's also Quen.
Bart:Yeah. We wouldn't have that without OpenAI. We wouldn't have that without Entropic. Right? Like, it's it's based on these learnings.
Bart:The question is a bit like how can you who is grabbing the application space more quickly if you don't look at give me a chat app. If you if you ignore that space because it's probably, today completely being owned by OpenAI. But if you look at, like, really industry or even at home appliances, it's it is probably China. Right? If you look at, for example and there's none that that is, of course, way broader than just AI.
Bart:But if you look at robotic installations in industry, if you look at the global world industry, 40% of installations are being done by China. Everybody else is small followers.
Murilo:Oh, wow. Didn't know that.
Bart:So it's also, like, they already have this huge bandwidth in things that are relevant where you can very easily inject AI going forward.
Murilo:But I'm also wondering, like, if you have to reproduce. So I think yeah. Like, I agree with you. Like, these two things are connected, and I think it's also a bit you're making a bet, let's say. But if you have to reproduce one because I'm also like, a lot of these things are not gonna be open.
Murilo:Right? That's a bit the idea. But if you have to reproduce one, is it easier to reproduce the the the integration of these things or to reproduce the the models? Not sure.
Bart:I think it depends a bit on the application space.
Murilo:True. True. Right?
Bart:That is a difficult thing. And then I think what China is simply very good at is manufacturing stuff and doing that in such a way that is super cost effective and scale up very quickly. And I think they are probably the best in the world of this. Like, if if Yeah. These are things where AI is relevant, they'll probably win that game in the short term.
Bart:Right?
Murilo:Yeah. Yeah. Yeah. Yeah. True.
Murilo:True. True. Well, to be seen. But and also the other like, I mean, maybe the last point I'll say make on this as well. Like, what is AGI?
Murilo:I think that's the thing that, to me, maybe we need to define a bit better because if we're always moving the needle, we'll never get there, which we have been moving the needle already. Right?
Bart:Yeah. And and maybe it's also a bit irrelevant in a sense. I think what what article basically tries to state is, like, in The US, and this is very black and white. Right? Like, put in The US, what they're doing is we need to get the next big major investment to have our next training cycle, to have even bigger data centers, and to have an even bigger model because that will be better than what we have today, and that that's the focus is very much, like, 80% on that and 20 on applications.
Bart:And while in China, it's probably the other way around.
Murilo:Yeah. That's true. Yeah. That's true. They they do make that point indeed, like, even make the the analogy with atomic atomic bombs and stuff in terms of power being generated and all that.
Murilo:Yeah. It's true. But I think yeah. Like, I mean, maybe I'm also being a bit biased because the article talks about, like, cure for cancer or military defense and all these things. But to me, it's a bit hard to to really see, like, okay, when do we get there?
Murilo:What's the next step? And what is this? And is there, like, is it really a risk that you're gonna get to a point where you're gonna say, okay, I'm here, and I have strategic advantage, or is it always gonna be something that you're gonna be chasing? Right?
Bart:If we only knew. Right?
Murilo:I think we will
Bart:get there at some point and then we will know.
Murilo:Maybe. Well, yeah, maybe. I don't know. Yeah. Yeah.
Bart:I think so. When you have your at home robot, slowly being depressed.
Murilo:You have to send it to, like, a robot therapies.
Bart:And to be talking about robots like this this this humanoid world games, they also made me think, like, we've been discussing, like, will there be job displacement because of things like Chechipee? Like, I think when you when you translate this to a physical world where people slash robots work with their hands, like, it's a it might even have a way, way bigger impact. Right?
Murilo:Yeah. That's true. That's true. I think yeah. Probably.
Murilo:Yeah. I feel like this is a sector of the people who are always safe or, like, stable. Right? But I think the more you have these things, the more
Bart:And the and the thing is just like there are a lot of things that we say, it's not possible to do this with AI. It's not possible to do this with robots, but everything is evolving so quickly. Like, today's not possible for a robot to do your plumbing, but, oh, twenty years twenty years is a lot of time. Right? We've seen major jumps in the last few years where
Murilo:they Yeah. Yeah. Yeah. Yeah. I think yeah.
Murilo:For sure. For sure. Like, if you look at these things as well, like the you just talked about the the humanoid Olympics or something.
Bart:Humanoid robot world games, something like that.
Murilo:Yeah. Yeah. Yeah. Yeah. It's like, if you're not that far.
Murilo:So
Bart:so It's it's actually worth having a look at, the highlights of the of the of the of those world games because at some point, I think it's during a 1,500 meter run, there is a robot that stumbles and it falls. Mhmm. And it loses an arm. And it's And
Murilo:it just keeps going.
Bart:And it scrambles back up again, and it and it starts going again. And you you hear audience going wild. Like, it's a very human reaction to a struggle of a robot, basically. Yeah. Wow.
Bart:It's a it's an interesting, experience. It's it's it's worth having a look at highlights.
Murilo:Hey. Do you know when you say struggle of a robot? It's very still. Like, it's it's anthropomizing. You know?
Murilo:It's
Bart:like Exactly.
Murilo:Exactly. It's like the guy's like, I'll do it. You know? The mission comes first. Alright.
Murilo:What else do we have?
Bart:Bear's creator, is changing the platform's license from LIT to Elastic after Copycat Forks threatens, the business. He says the new terms are almost the same, but ban running bear as a hosted service aimed at stopping free ride competition. For developers, it's a nudge to balance openness with sustainability and to expect more source available moves as AI speeds up cloning. So bear is short for bear block. It's a it's a blocking platform.
Bart:I think it's the best one out there. You can you can test their bearblock dot it's the one where I host my personal blog as well. Bart space? Sorry? Bart dot space.
Bart:Yeah. Yeah. And they were they they were on the on the MIT license, and they moved to to Elastix one. I'm I'm I'm forgetting what forgetting what the name of the Elastix one is exactly. Let me just see what no.
Bart:I don't have the exact name of the Elastic license. But it basically says that, it's it's more or less this Elastic license is more or less a sort of a copy of the MIT license, but it basically states that you're not allowed to run this as a service. So it's as simple as that.
Murilo:I think it's called Elastic license. Like, it's from Elastic. They created a license.
Bart:Yep. And it's just Elastic. And he he's he's stating that he's done this because he noticed that there were some copycats, basically, office work that that did no back contributions and basically were were onboarding users and basically missing revenue for him. And I think this is also a good example of a very, like, very personal project. It's one it's a one man show.
Bart:He's very super passionate about what he's building. He's always focused on open source, and it's just painful to see your work being taken and then not really as a learning point, but just to to do exactly the same and have someone run it and just basically take your customers from you. Very hard earned customers. Indeed.
Murilo:Also, the he also comments on with agentic AI. Right? AI assisted coding. It's even easier. You're just, like, go add some stuff, change some teams and stuff, and then boom.
Bart:Exactly. Exactly. So
Murilo:yeah.
Bart:To me, the the the to me, it's very understandable what he's doing. I was at the same time a bit wondering, like, because you have generic coding, like, it also becomes like, even if it's an Elastic's license, it is somewhat easy to circumvent this. And maybe it's not perfect today, but it will probably be perfect in two years where you say, okay. Look at this source available repo. I want something like this.
Bart:Needs to be it needs to look different. The code can't can't look like it, but it needs to have all the same features. Yeah. And maybe you built it in a different stack. Right?
Bart:Like, no one will see anymore that it came from that code, but it's like very, very minimal effort to build something off of something that is source available.
Murilo:And it's also and it's If I looked at a code and I really just I mean, I can imagine now that the AI is not AI. It's actually a person. I actually read the code and I have the same reflex like, ah, this is really cool. I have some really interesting features, but I don't like the stack. I'm gonna do my own stack, and I'm gonna build it from scratch.
Murilo:Would that be violating? You know?
Bart:No. But I think to do that, it's a bit it it is maybe the same thing, but it feels different. Right? Like, if I if I'm if I wouldn't use generic coding, then I look at this source available, and I'm I'm super passionate about building a blocking platform. But I'm opinionated because I want to do this in a different stack.
Bart:But I do care about the features that Herman, he's called, has built there for BareBlock, and I'm trying to understand how we did it and and and to learn from that. And based on that, put a lot of time and effort to make my own platform Yeah. Which in the end is exactly the same, but it feels different than just saying to Jenny, I look at this, learn from this. I want exactly the same features between this different stack. And then after an hour, have something that is up and running.
Murilo:Then it's really just about the effort. Right? Because I think on the difference between the two is that one of them is Or ethics. Effort. Yeah.
Murilo:I think but I think to me, the ethics a bit the same on both. They're both equally unethical. I think the thing the main thing the main difference for them between them for me is just that one, you're not putting any effort. You're really
Bart:Yeah. That to me, that is the difference. Yeah.
Murilo:Yeah. Yeah. That's the that's the difference. Right? Like, in one But
Bart:I don't think there's like, in the first example, I don't think there's anything against being inspired and learning from an open source or a source available project.
Murilo:Yeah. Yeah. I agree. I agree. I agree.
Murilo:I don't think there's anything I don't think there's anything unethical about that. I mean, I think that's why you also make the code. I think David mentioned, I think. That's why it was source available.
Bart:Yeah. Yeah. Right? It was even just MIT license up to a very short. It's it's sad that this needs to happen, but it's understandable.
Murilo:Yeah. Yeah. Yeah. But I think it sparks the debate of, like, open source sustainability and all these things. Right?
Murilo:Like, how how can we still support? I think the world depends on this. So how can we support the the open source projects of the world? Right? Maybe one last question.
Murilo:Why is why is Bear your favorite blogging platform, Bart?
Bart:I think it's a very fair price, Tom. I think it's also I like without having the good arguments for it, but I like these minimalist type of blogs. It's very there are very few frills. It's just markdown that you type. It's super easy to configure.
Bart:It's it's opinionated enough to just work well just out of the box. I'm very happy about it.
Murilo:Okay. And there is you're you're hosted on you're hosted there or no?
Bart:It's a it's a hosted service. Yeah.
Murilo:Yeah. It's a hosted service.
Bart:You pay us by a very small monthly fee. I want to say $3, but I'm not sure.
Murilo:Cool. Cool. So or people check it out and support the project. Right? Yeah.
Murilo:What else? Anything else you wanna say on this one, Bart?
Bart:Nope. Let's continue.
Murilo:Let's continue. So what else do we have? We have Cloudflare dug into AI crawler patterns and found training drives most bot traffic while referrals back to publishers are tiny. In one snapshot, Anthropix crawl to refer ratio hit 5,000 to one, OpenAI, 870 887 to one, and Perplexi, a 118 to one. Hence, the blunt line, no click through, no eyeballs, and no ad revenue.
Murilo:If you publish on the web, this means tighter bot controls, clear content signals, and new rules for how your work gets used. So this is from Cloudflare. I think we had an article before on Cloudflare and how AI bots crawling their websites actually changed a bit the game. Right? And maybe a quick recap from that is the now opening, I can actually look at your site and crawl it once, which doesn't give you value because there's no ad.
Murilo:There's no one buying stuff. And then you can just instead of directing people to your website, you can just kinda give you the the answer, give you the content without necessarily giving you the traffic and the ad and the marketing, quote unquote. Right? Mhmm. And, actually, I think the article the article is also about it was also from Cloudflare, I wanna say.
Murilo:So what is this one about, Bart?
Bart:Yeah. Cloudflare has been focusing a lot on on AI crawl crawlers recently. So, like like you said, they also they already did some reports. They've also added some features. I actually moved my DNS configuration for a domain to Cloudflare today.
Bart:And it actually, while, configuring it, immediately asks if it needs to block AI, crawlers or not, if it needs to, set up a robot c x t specifically to instruct crawlers. So it's doing a lot around this. This specific article goes into the new insights that they have around AI crawlers, and there is if you look at the second link, Milo, in our in our notes, it's the AI insights Cloudflare radar. And there are a lot of a lot of things that you can learn from that. Like, there's really, like, how much HTTP traffic is there per bot?
Bart:What is the crawl purpose of these bots? What is the crawl to refer ratio? Like, how many times do they crawl, versus how many times do they actually send someone to your page? There is also I found it very interesting. There are AI bot best practices.
Bart:Practices. So there's also there's, for example, it shows, like, is this bot verified by IP? Does it use distinct bots by purse purpose? So is it a distinct bot for training or crawling, etcetera? Does it respect robots.
Bart:Txt? And so, for example, to zoom in just there, the with these best practices, we always have this I think in the community, it sort of lifts that, like, Anthropic is the good player.
Murilo:Mhmm.
Bart:Right? But if you just look at the crawl to refer ratio, Anthropic is by far Yep. The the biggest abuser. Like, this is, like, 33,700, crawls for one referral. For example, the the the the next biggest one is OpenAI, which is a fraction of that.
Bart:It's 908, still a lot, to one referral. But it's a fraction of Anthropix. And, also, if you look at the AI boss best practices, almost all the big ones, like, they're, example, verified. Their bots are verified via IP. Anthropix one is not.
Bart:Doesn't respect robots. Txt. Almost all of them respected Anthropic. It's not clear if they do. Right?
Bart:Like, it's at at least, it it this shows a bit of a different picture than that we have in the, quote, unquote, community on. Yeah.
Murilo:It's it's true. It's true. It's a bit it almost makes me a bit sad because it's, your favorite. You know? You're a good guy, then it's, like, he's not perfect.
Murilo:Like
Bart:And what also surprised me, there is there is this this small graph on generative AI, the popularity of services. I think the the the number one doesn't really surprise anyone is ChatGPT. And then the number two, it's it goes either between it depends a bit on the month or on the on the day. Goes either between Anthropix's clot or Character AI. I have I personally had no clue that Character AI was this big.
Murilo:I don't even know what character I never heard this.
Bart:Character AI is like your your virtual friends.
Murilo:It can
Bart:even be your virtual girlfriend, Mila. But, like, these are these virtual avatars that you can talk to, basically, and that can also crawl the web for you. But I had no idea whatsoever that it was that
Murilo:big. Wow. Yeah. And this is the yeah. This is an interesting thing.
Murilo:Gives a science science backed. Right? Like, it's not, like, it's not vibes.
Bart:Yeah. Wow. So interesting stats. So definitely visit their AI insights at Cloudflare Radar and check it out yourself.
Murilo:Indeed. Maybe a few questions for you, Bart. You you switched DNS to Cloudflare. Why why does one switch DNS providers?
Bart:Because I use Cloudflare for all my DNS management because it's very easy, and it's a bit of a central hub. Oh, okay. So you do switch. It's always easy to switch your names name service to Cloudflare so you can configure your your DNS there. But I don't always buy my domain at Cloudflare.
Murilo:I see. So you recently bought a new domain, and then you moved to Cloudflare.
Bart:Exactly. I I manage the DNS via Cloudflare.
Murilo:Yeah. And then which domain did you buy?
Bart:For me to know, for you to find out.
Murilo:Okay. Okay. And the Cloudflare, did you so when you moved, do you notice these new features about AI? Like, about the AIware?
Bart:Like, few months ago. When you when you configured a new domain.
Murilo:Okay. Okay. And then for the old ones, you still need to go there and manually change the things, I guess.
Bart:I guess. Yeah. It'd be interesting to see what they did with the default settings, but I I doubt that they switched to default blocking. I think that would be weird.
Murilo:Yeah. Yeah. Yeah. It'd be very so it could be very
Bart:disruptive. Well, there's also I mean, people also need to keep this in the back of their mind. There is a monetary incentive here. What they want to be they are, for a big part of the Internet, they are a bit the the gateway slash gatekeeper. What they want to do is at some point allow, website or domain owners to to basically ask a fee for crawlers to crawl them.
Bart:So there's also this monetary incentive of them to say, oh, wow. Look at how big this this crawl versus refer issue is, and you should really do something about it. So I think people should not forget about that part either. Like, there's something in it for them to Yeah. To point this out as well.
Bart:But it's good that we have some actual statistics on this now.
Murilo:I think so too. I think this is a I think it's good that they're it seems that they're very active on this debate, and I think this is something very disruptive. And I think ad revenue on the web is a is a big thing, and I think AI is very disruptive there. So I think it's good that they're giving options. They're providing.
Murilo:They're educating as well. That's something I I hadn't thought about before, but then I started hearing more and more. I was like, yeah, this makes a lot of sense. Right? So so yeah.
Murilo:Interesting. Cool. And I like to see the the data in here to see how this whole thing is gonna unfold as well. A lot of businesses, a lot of people's, a lot of jobs are related to this as well, so it's all gonna change. Who else's job is gonna change Bart?
Bart:Aspiring artists are using AI tools like Suno and Yudio to launch careers, sometimes landing record deals or millions of streams. Platforms and labels are still wrestling over licensing and credit. Even a services say a chunk of upload, diesels these are packs at around 18%. AI made. For listeners and creators, the takeaway is simple.
Bart:Great IDs now matter more than gear, but the legal lines are
Murilo:This one is AI music?
Bart:This is AI music. This is we have some numbers. I was wondering how they get the numbers. It's not that easy, I think. But these are the if if if you know, you know.
Bart:These are a lot bigger in the past, but it's a bit of a Spotify alternative. They estimate that 18%, 18% of all new music uploads is AI generated, which is a lot. Right? Like, mean, that's two and ten.
Murilo:That's lot. Even though it's probably, like easy
Bart:to recognize it. Things are hard to recognize it, So I wonder how accurate numbers are, but let's I mean, still, if it's 5% less or 5% more, still make major chunk. Yeah? I I don't know what it means for the music space.
Murilo:Yeah. I I don't know either. Think I've always said that music or art, lot of times, is about how it makes you feel. And I think a lot of the times, by knowing that it comes from a person, it makes you feel differently than if it came from a machine. But I'm not sure.
Murilo:Like, I don't know. I feel like a lot of the times the line is not so clear. Right? Just because the machine created the the course or something, maybe some a human actually curated it. Right?
Bart:Yeah. And especially if you compare this to something like which is already a lot of computer behind it, like something like ETM.
Murilo:Yeah. Exactly.
Bart:Right? Where there's a lot of like, probably 80% of what, what comes out has been has been digitally produced. Yeah. And, also, if you compare it to something like like hip hop beats where traditionally you see a lot of parts of those beat being sampled from all the records, which is a bit like the commentary on these these these I music generator. It's like you're using our material to do something.
Murilo:Yeah. That's true.
Bart:You see this in other music genres. Well, I think the difference is, like, if you look at hip hop, for example, like, if you use a sample from something where there's a copyright, you need to clear the sample first. So you probably you're you're paying the original artist, basically. And I think that is a legal line that you don't have here because probably the original sample does a lot to the output, but it's not exactly that sample anymore. Right?
Murilo:Yeah. And it's hard to know exactly how much each sample contributed to this output. Yeah. Yeah. Right.
Murilo:I do think that in the heart of it, though, it still is about people are finding the easy way without the effort. And I think that's the thing that rubs people wrong in a lot of ways. Like, you should be rewarded for the effort you put in. Yeah. This guy, like, the guy in this picture here, Oliver McKen, he does say, like, I have no music talent at all.
Murilo:I can't sing. I can't play instruments. I have no musical background at all. And he has 3,000,000 streams on some platform, and he actually have signed a record label, yeah, with Hollywood Media.
Bart:That's crazy. Yeah.
Murilo:And I think I think I mean, it's it's insane. I mean, in the article, they also talk about, like, how people, they they still write the lyrics sometimes, but then they have like, they spend hours, like, nine hours to prompt the AI to get to do what they want. So maybe it is done a lot of effort still, but it's a different type of effort. But then I think I think I think that the thing that sits weird is, like, the fact that people that don't put the effort to really learn the skill to really hone it, now they're having the same benefits in a way. Right?
Bart:Yeah. I'm think when it comes to this, comes to this aspect at least, I'm a not to the copyright aspect, but to this aspect, I think I'm an AI optimist where I see this more as a tool to make the analogy again, like, with EDM or making beats, like, where you like, you have a lot of tools in this space. Take, for example, Apple Studio, which is very big in the hip hop scene. It's actually by a Belgian company, Image Line. Probably the biggest production tool in that in the hip hop space.
Bart:And you have people that can't play any musical instrument, that can't sing, but are great, like, extraordinary good at making good beats.
Murilo:Yeah.
Bart:And the only major difference here is that it's a completely different tool. And, it's a tool, of course, that came on very quickly and is a bit of a shock to the industry. But to me, it's not that different to something like Apple Studio.
Murilo:I agree. I I actually and and that's the the second thought I usually have. I agree with what you're saying. I also think that at the time, a lot of the electronic music as well was perceived the same way for similar reasons. Right?
Murilo:Like, you don't don't play the guitar. You have no skill. Now you're just pressing buttons like DJing. It's like, what are you doing? You know?
Murilo:So I I agree what you're saying. I think it's a very different tool, but I think it's still a tool, but that doesn't like, like you said, doesn't exclude the the copyright thing. Right? Like, maybe these tools were only possible because they infringed a lot of copyright laws.
Bart:I think that is that is still a very difficult discussion, and it's a bit like me listening to an old record and being inspired by that by that, but that's times times a billion.
Murilo:Yeah. Exactly.
Bart:And you just get it out of the box. Yeah. And that feels unethical.
Murilo:And Yeah. And
Bart:Honestly, like, with the lawsuits that were that we see in the first the first conclusions around that, We had a recent one on Entropic reading all these books or using using a huge amount of books, basically, the the LipGen and LipZealip books book set where the they ruled that, basically, reading them all was fair use, but obtain obtaining them in a in legal manner, that is a problem.
Murilo:Yeah. Yeah. I remember the we discussed this as well some time ago. Right? Like, the the using the knowledge is okay, but the acquisition of that knowledge is not okay.
Bart:Yeah. And that's I mean, if you translate it to this space, like, that basically means that Suno or Yudio, like, they should basically do one stream for every song that they that they use in their training set. But that, like, economics wise, it still doesn't feel fair. Right? Like, I just do one stream of everything.
Bart:Like, it doesn't benefit one of these single artists at all. Like, the the benefits are marginally almost zero. Right?
Murilo:Yeah. You
Bart:just do one stream. So
Murilo:the Exactly.
Bart:Yeah. Still I I hope it's something that we see evolution in.
Murilo:Yeah. A better a better outcome, right, for the for the music producers because indeed, yeah, it doesn't feel like the the balance is is right. Yeah. It's right there. What else do we have?
Murilo:We have AI influencers. Financial Times charts the rise of the AI influencers. Virtual personalities that brands can spin up fast, cheap, and endlessly on message. The piece notes that, and I quote, computer generated profiles are gaining traction with brands but may pose a threat to real life endorser, end quote, with early campaigns already blurring the lines. If synthetic create greater scale, marketers may save money, but audiences will demand clear labels, and humans will have to lean on into authenticity.
Murilo:So what is this this one about Bart?
Bart:So the Financial Time, basically, they report that, more and more and more brands are basically spinning up these CGI slash synthetic personas that are very fast and are very cheap, and they're basically always on. They're always ready to message you back when you send them a DM. And it's, it's starting to also, like, raise displacement and transparency concerns for, like, versus basically human creators. A little bit of numbers here. Like, the the the numbers are they're mentioning how big the space already is.
Bart:Like, they estimate the full space to be worth today. Well, it was 2024 number, 6,000,000,000. They estimate it to be 45,000,000,000 by 2030, so it's a very, very fast moving markets. Do also have some, let's say, early signals that is also that this not only exists, but also is generating profits, basically. We have a Spanish a Spanish virtual person that apparently earns up to €10,000 a month, as a as a virtual influencer Wow.
Bart:Basically from brand work.
Murilo:And maybe this they must
Bart:say still outperform on engagement.
Murilo:Yeah. And when but when you say, like, a virtual like, a Spanish AI influencer, is it a the creator of that influencer is a Spanish person?
Bart:No. It's a I think their persona is a Spanish persona.
Murilo:Ah, the persona is Spanish. Yeah. Yeah. Okay. So you don't know actually where?
Murilo:Sorry? Like, you don't know exactly where the AI is created, managed, or or anything? Exactly.
Bart:Okay. Yeah. It's I think that's a difficult thing. Like, it also, like, it's creates a bit of questions about authenticity when you see these type of things. And I and I saw this article passing by, and the reason why I actually pushed it in here as one of our one of our things to discuss is because I noticed very much when I am on Instagram, but also on TikTok that there is a lot of AI generated content.
Murilo:Yeah.
Bart:But, like, a lot. And it's and I think it's it really boomed after what's the name of the model? The Genie three model. Am I correct? No.
Bart:No. That's three
Murilo:world the world generations.
Bart:Yeah. Yeah. But the the video generation model of, of, Google that also has audio generation.
Murilo:Video three or something?
Bart:Yeah. Thank you. Thank you. After that one, it really started to boom, and I think I don't think it's it's a bit now I'm just doing from my specific bubble. I don't think it's an overestimation to say that 10% is AI generated of videos
Murilo:I don't think basically. Yeah. I don't think so. I think there's also there's some clearly AI generated ones. There's, like, with the animations and stuff or with the audio, you know, like, you can tell.
Bart:Yeah. Well, you say you can tell, but I was the other day, I was sitting next to someone. I said, hey. Look at this. And and and the guy was saying, but what's wrong with it?
Bart:But they say, it's AI generated. This is the bullshit. And he hadn't he didn't he didn't see us.
Murilo:At
Bart:all. And I think we also because we are very active in the space, we overestimate how easy it is to spot this.
Murilo:Yeah. Maybe. We also I mean, we see it evolving. Right? And we look for the the queues for, like, this is AI generated.
Murilo:And, yeah, for people following, this is the the Google v o three model. I think this is the one you're saying. Right? Very, very
Bart:And also, but but you see at the same time, it's I think it's also a trend of the last year, maybe, Trump influenced or not, is that there is much less focus on let's make sure that this is safe. Let's make sure that we can't generate deep fakes. Like, it's become very easy to make deep fakes even with, like, the new new image, manipulation model now by Google. Like, it's very easy to generate famous people next to you in the picture. Like, like, all these things become easier.
Bart:So it's a very slippery slope, I think.
Murilo:Yeah. That's true. That's true. I think, yeah, another thing to to look out for in a lot of ways. Right?
Murilo:And I think it's another area that I think people, and I think kids need to be educated. And I think we talked about this in the past, how teachers now, they're gonna be in a place where they have to teach kids skills that they never learned, Right? Like, as kids as well. Like, how do you look for artifacts on these videos to see that it's AI generated? How do you know about fake news?
Murilo:How do you verify information? Right? So, yeah, I think let's see. If if is listening or hearing me say this on our shorts, this is not yeah. I just hate it.
Murilo:Just to be clear.
Bart:I can vouch for you. You really Yeah.
Murilo:Unless we're both
Bart:Unless we're both. Alright. Simulate the world. Maybe everyone there.
Murilo:Maybe everyone except me. Yeah. What else do we have on that note?
Bart:We have Is it my turn?
Murilo:No. It's your turn. It's your turn. Go for it.
Bart:My turn. My bad. With Python Python, the documentary. Cult Repo's new feature length film tells Python's origin story through the people who built it. From early experiments to global ubiquity premiered on August 28 with a live YouTube chat, the about eighty five minute documentary threads, interviews, community milestones, and how Python spread well beyond the web.
Bart:For tech history fans and teams hiring Python talent, it's a crisp primer on why this language keeps winning. I don't have too much to say on this. I think for anybody that uses Python actively, I think it's very much worth a watch. I watched it two days ago.
Murilo:You liked it? There were a lot of sorry? You liked it.
Bart:I liked it. Yeah. I think it's, it's very much worth to watch. There were even though I've used Python for a very long time, I've actually had the first O'Reilly book about Python three. I bought it immediately when I came.
Bart:Really? So, yeah, I I was there early days Python two. Today is a very long time. But the even the story before that, how it came to be, how originally it was not gonna be Python, but it was gonna be the ABC program and language. I thought for a very long time, actually, Guido van Rossum was at CWI, but it was it was for a very much shorter period than I thought.
Bart:There's also discussion on how the Python organization software organization. What's it called again?
Murilo:PSF, Python Software Foundation.
Bart:Thank you. How the how it came to be, also how the company behind ZOOP, which was at one point a very big Python framework, is probably one of the key drivers on why Python today is even still there. Wow. Stepped up at a very crucial time. So it's really worth to watch.
Bart:Yeah. And also I what I also thought was interesting, I think most people involved in the Python community, they also know about PyLadies. It's actually that played a very active role in making sure that there were women also poor part of the core developers.
Murilo:You know,
Bart:the person that's that mentored the first woman core developer himself. It's a it's a it's very interesting watch.
Murilo:Very cool.
Bart:Yeah. It makes me like the PiDA community even more.
Murilo:Nice. That's nice. I heard that this I think the trailer came out on, PiCon. Yeah. I'm I'm curious.
Murilo:I'll I'll definitely give it a watch. I think it's always always interesting as well to see how these things come to be. Right? I think, again, everything's a bit of a rabbit hole, and I think to see how something gets built, something gets shaped, to see, like, the the things that go beyond the programming. Right?
Murilo:There's, like, the organizational part and all these things. So I'm very curious for sure. I'll definitely give it a give it a watch, then we can chat about it more. Alrighty. Last but not least, Stanford researchers analyzed payroll records and find early career workers in AI exposed jobs are already bearing the burn the brunt of change.
Murilo:Since generative AI took off, and I quote, early career workers aged 22 to 25 in almost AI exposed occupations have experienced a 13% relative decline in employment, end quote, with cuts clustered where AI automates tasks. If you're just starting out in exposed fields, the signal is to upskill toward augmentation and chase roles where human judgment compounds. So this is a paper, right, from Stanford.
Bart:Mhmm.
Murilo:I didn't actually read the whole paper.
Bart:I just It was a paper. Yeah.
Murilo:Yeah. But I think he talks a bit on something we speculate a bit. Right? Like, how AI jobs how jobs are affected by the rise of AI.
Bart:Yeah. We actually felt it a bit already last year that, it was harder for junior people just graduating to, to find a job. This is one of the bigger, research papers looking directly into this thing, like like, what is the impact of AI on on whether or not there is job loss. It's a bit of a doom story, maybe. Yeah.
Bart:Because not a positive one. So the even though, apparently, and I'm a bit paraphrasing here, like, the overall jobs seem to be up. But where there is exposure to AI, and then there is a bit of a difference between automation and augmentation. I'll come to that later. But where there is exposure to AI, young workers are basically lagging.
Bart:Right? And you mentioned the the 30% decline in in young aspiring people headcount. When it comes to specifically software developers, it's even down around 20% from its 2022 peak. 2022 was of of course. Post COVID was also very, very high.
Bart:But still, they also like, they explained in the paper how they try also to correct for these COVID effects for system shocks. And even when they correct for us, apparently, they you still see the same pattern where overall jobs are up, but junior people are having a harder time finding a job. They also make this comparison between automation versus augmentation, which, of course, is a bit of a nuanced one. Right? Yeah.
Bart:But it comes from Anthropic's economic index, something that we also discussed a while back where they make this definition. And where you see that where know, if you're in a field where AI can cause automation, you really see a large decline in junior numbers when they're when they're in a field that where there where there's augmentation, you don't see this impact, basically.
Murilo:And I think So basically, like, replacing reduces jobs and augmenting, you don't see. Even though I would expect also decrease now. Cause like if you are augmented by, and I'm thinking really product productivity, if you're augmented by AI, you probably can be more productive. So one person can do the job of three people now. So either you have more job or you have less people needed.
Bart:So, yeah, just to give some concrete examples here. So the one they what is automated heavy, they mentioned, for example, accountants and auditors, really analysis, like, in a very structured manner of, for example, financial numbers, but also for developers, like, we've just discussed the discussed the AI generated coding a lot, information clerks. When it comes to so the this this type of examples are apparently have a big impact when it comes to, where there is basically a lack of junior employment with augmentative, examples. This, for example, and this is very varied. There's there's maintenance and repair workers that are able to do their work more efficiently probably so they can do more, I guess, or better quality.
Bart:There's also this example of registered nurses that are probably using something like chest GPT to to, to up their knowledge or how I don't know how how this translate exactly to the day to day work. There's also the example of, like, if you're a computer or information system manager or an architect, which, by the way, is probably not really a junior role, but like, it also, like, it empowers you. Right? Like, it empowers you in that role, something like JTPBT. So yeah.
Bart:And and I I I also have this feeling like these examples, like, heavy versus augment heavy. It's also like it depends on what is the state of the technology today. Yeah. For sure. I mean, take it to the extreme, a registered nurse if robots are perfect Yeah.
Bart:With our discussion of earlier, like, maybe that also becomes automated heavy, that field. Right?
Murilo:True. True. I think I'm also wondering I mean, you're talking about nursing, there's the physical aspect, but there's also the the procedural more. Right? Like, someone has this condition, push this medication, person has this reaction, push this, which in a lot of ways, I think maybe the the maybe not AI like LLMs.
Murilo:Right? But, like, a artificial system could already be more efficient. Mhmm. Yeah. In the sense of, like, the procedural knowledge is always there.
Murilo:There's no fault. There's always things by the book. Right? There's no panic. There's nothing.
Murilo:So which I also think even with medical, like doctors, medical imaging diagnosis, right, I think AI has also a potential to disrupt those, which I also know that caught a lot of people by surprise because there was also the assumption that the not highly intellectual jobs are not going to be the ones that are going be replaced by AI or by automation in general. Right? Which is not what we're seeing. Maybe this does not have it hasn't really become true yet, but, like, we see potential there. We see chatter.
Murilo:We see research, right, that puts those jobs at risk.
Bart:Yeah. And and and what is also mentioned in in these articles, like, we have a lot of different trends playing at the same time. Right? Like, this is we have we're in a post COVID where there was a lot of hiring immediately after COVID. You're also in this space where economic or geopolitical stability is very, few and far in between.
Bart:So it's hard. Like, I think all of these kind of thing also make companies hesitant to hire. It's I think it's easier to postpone junior hires even though you might have a issue in the long run then.
Murilo:Yeah. I think yeah. I agree. I think, again, we talked about it, like, last week, I wanna say. Mhmm.
Murilo:I think junior hires are also the ones that we take more for granted because if you miss this train, the next train will come Yep. Later. Right? So so yeah. But
Bart:what this article is is trying to say, and they corrected a lot for these trends like that, these juniors might be the the canaries in the coal mine where you where you basically see that, like, the canaries in the coal mine for people that don't know, like like, back in the day. Well, there's still coal mines around the world. Right? But we don't have any any here. We're not canaries.
Bart:Back in the day, they used the canaries and a small cage in the coal mine. And when they died, people made sure to get the hell out because that was a sign of carbon monoxide. And here, like like, these early career workers, like these juniors, like, they might be a sign, like, something is wrong. Like, we need to do something because jobs are down for them, wages are flat. Like, something is happening.
Bart:Right? Maybe we should not close our eyes to this.
Murilo:We shouldn't wait and say everything will be fine. Yeah. Yeah. True. True.
Murilo:And do you think do you agree with the overall sentiment then that we should do something?
Bart:I think it's very hard to say what you should do. I think that is a difficult thing. And this goes like because what you've you've so many discussions on this. I think the extreme is no. We just need to move still, like, go very far quickly ahead so that we know we can do it.
Bart:Let's fix it after. You have voices, like, aiming to regulate and basically slow down until we've tested the waters, and then we have some guardrails in place. And you've other extremes, like, where, like, where it's okay. Shit's gonna hit the fan anyway. We need something like universal basic income.
Bart:I I think the difficult thing of all of them, they all have a lot of pros and a lot of cons.
Murilo:Yeah.
Bart:Right? Yeah.
Murilo:But I do think Or
Bart:at the very least, it's very hard for a large part of the population. And in this case, it's probably not a large part of couple of population of a country, but of the world Yeah. To stand behind any one of them. Think that is a difficult thing here. Because if we say maybe Europe is not a good example because we're slow as fuck on this.
Bart:But if the if we if we would say that we we just let's slow down. Let's regulate everything. Let's let's make sure that we take one step at a time. Like, we're completely outpaced, like, on every side. Right?
Murilo:Yeah. No. That is true. That is true. But I do think at least I do think we are paying attention.
Murilo:I think we are like, we I'm saying we as the world, right, like, are asking the right questions as well. So I think we're not at least we're not gonna be caught by surprise. Right? I think, like like you said, I don't know what's the right answer, but I think there's gonna be many different answers. And I think we're also gonna learn from each other, and I think we'll find a solution.
Murilo:I don't know what it is, but I think we'll find a solution. It just might be my optimist.
Bart:Let's let's be optimistic.
Murilo:My optimistic side there.
Bart:There's enough in the world to be gloomy about already. Right?
Murilo:Exactly. Alrighty. And I think this is this is all the topics that we had for this week. Anything else you wanna say before we part ways, Bart, before we call
Bart:it apart? But it, it makes me think what I would like to discuss at some point is what's your Europe's future gonna be?
Murilo:My future.
Bart:No. Not Europe. My old is my my my accent. What is Europe's the EU's future gonna be in EU. In AI.
Bart:I'm a bit pessimistic, but we see also major initiatives. Like, we have sort of the similar similar example to the CHIPS Act in The US. We have a forgot the exact abbreviations, a long abbreviation in the EU. But you have initiatives coming out of this, like OpenShift that are trying to basically, build AI accelerators, so really on the hardware side, but I'm a bit skeptical on on the the future. It would be interesting to talk to some people on that and deep dive on that.
Murilo:I agree. I think I'd honest, I personally don't feel like I have enough knowledge or information to comment much on this. I think most of what I would say would be speculation, but I do share the concern. Right? I think because this is the kind of things as well that you start to fall behind, and then the gap just the the the the margin just widens.
Murilo:Right?
Bart:And Exactly.
Murilo:Yeah. It's not something that you if you just wait, you catch up, and that's fine. Right? So I I shared the concern. It will be interesting to talk to more people.
Murilo:And what are the main differences and what are some solutions being proposed? Right? I think, of course, the the EU's values are very different from what US is. And, like, is there a future where we can kinda keep the values and and still move fast and still not fall behind? Right?
Murilo:I think that's the and I see there's a lot of people thinking about these things as well. So Mhmm.
Bart:So it would be cool for sure.
Murilo:If you know someone, if someone listening knows someone, let us know. Be very, very curious. And that's it. Any anything else you wanna say before calling before wrapping it up?
Bart:Thanks for hanging with me, Mugula. Thanks for listening to all of you out there.
Murilo:Yes. Thanks, everyone. And see you next week.
Bart:Listen. See you all next time.
Murilo:Yes. Ciao. Ciao.
Creators and Guests
