OpenAI tiene un arma inesperada para acabar con Anthropic: portales extraterrestres

Hola a todos los monos estocásticos y bienvenidos al podcast favorito de los corredores de media maratón de Pekín concretamente, porque ahora nos escucharán mientras son adelantados por robots bípedos, los de Honor y este tipo de compañías. Así que un saludo y un fuerte abrazo para ellos. Antonio Ortiz, ¿cómo estás? Andaba pensando yo, Mati, en esta intro comentar mis cositas de Runner. Because the runners lent now we are vindicated. Atención. I, of course, you and I invented the hybrid sport, the hybrid sport. And, of course, we use the race, but to go slow, to go in zone 2. If we are técnicos. So, of course, always the smile of the runner quick that goes on the pace of maritime, What happens to your veras, dando grandes zancadas e imprimiendo, además, una especial velocidad cuando te sobrepasa. Pero, ah, quien ríe el último ríe mejor, Mati, porque ahora los runners de lento estamos reivindicados, porque, ¿qué más da? Corre más rápido si al final un robótico es el que va a hacer el récord del mundo. Pues mira, te ha pasado lo que a mí ya me ha pasado como periodista, ¿no? ¿Para qué escribir si los modelos de lenguaje te pueden escribir 200 articles al día y a lo mejor lo hacen mejor que yo. Pues los runners, como tú dices, Antonio, solo sobrevivirán los que no estén ahí para hacer el mejor tiempo porque el mejor tiempo ahora se lo van a llevar los robóticos. ¿Novedades en el mundo de la inteligencia artificial? Pues tenemos muchas cositas hoy, Mati, en el menú. Tenemos muchas propuestas, lanzamientos and debates, but I think we have to start with what we have in the warmest image model, the image 2 integrated in chat GPT. For finally, competition for Nano Banana, because Nano Banana Pro and Nano Banana 2 Google llevan bastantes meses y continuando también el legado de la primera versión de Nano Banana, reinando en el mundo, en el reino de los modelos de imagen, simplemente por la versatilidad que tienen. Los modelos de Google de imagen son inteligentes, entienden lo que le estás pidiendo, razonan antes de generar la imagen. le puedes introducir imágenes de referencia y te va a sacar el mejor resultado del mercado. Y no tenían competencia hasta que por fin OpenAI ha dicho aquí tenéis Chuck GPD Image 2.0 esto es otra cosa, es otro nivel y ahora cuidado con las capturas de pantalla que circulen por redes sociales porque a lo mejor no son capturas de pantalla a lo mejor es una ficción construida con ChagPT. Sí, además, bueno, me voy a saltar todo lo que tenemos en el guión relativo a las capacidades, los Bessmann, etcétera, porque hay un caso que creo que va a afectar mucho a la ciudadanía, que es el de la presentación de capturas como pruebas en juicios. Tengo un amigo recientemente que se ha visto involucrado in a judicial dispute, and he presented capturas of WhatsApp as a probator, as a proof of to convince the juez that he was right in the litigio. Well, I don't have much experience, but basically, it's recommended to have a certain operative, someone, a expert in informatics, that can tell you that that that is not an inventive, that is not a modification, And I think this step of the capture per se can be taken as a test, I think it's going to take a consideration in these times because the level that I have seen with this new image image image of the GPT, a lot of the time, the picture of the screen, the chat, WhatsApp, or any other type of digital that circulates and has been able to do the training. It's incredible. The level of this model has surprised me for very well the capacity that it has to follow the instructions and the capacity that it has to work with details. What we did in Echo En el último episodio que hicimos en entrevista sobre por dónde se mueven los modelos creativos en imagen, en vídeo, en esos dos campos entre tener mucho más control sobre lo que va a salir en el detalle, el que refleja exactamente lo que tú tienes en tu cabeza y que le has descrito, o el potencial de hacer un montón de cosas diferentes, muchos estilos, ser más creativo. I think this model, as you say, is going to be a banana, to try to push in two ways. And my impression is that it's put the number one. My impression is that it's put the number one, without a doubt, with one thing very a favor and one thing that I'm going to do. The thing very a favor is that it's put in the chat GPT in the best way of approaching the reasoning of the best models of OpenAI. So when you ask something, you can think and think and find a plan on how to be the image. The monos estocastic, we share it in our profile in X, when basically the instruction that I was giving you to the model there was, well, Well, it's the YouTube channel of monos-stocastic, if we change the editorial of the artificial intelligence to the pistachio, the pelvic floor and the microbiota. So basically he does a razonamiento, entre comillas, a exploración, a planteamiento de cómo serían las portadas de esos videos, de cuáles serían sus títulos y todo eso lo discurre el modelo. No es algo que básicamente yo haya tenido que describirle paso a paso. Entonces en esta parte me parece que es un modelo chulísimo, que realmente es lo mejor que hay ahora mismo. but in the creative part of making works, images a little more evocative or with a capacity or with a higher level of abstraction than the literality of this, then there is a impression that at least in my first test I would put a little behind other models like the Flux or the same Mick Jarney, which is always a little bit un olvidado en todo esto, pero que ahí siguen peleando. Y bueno, esas son mis impresiones iniciales. Yo creo que si los modelos de imagen siguen por esta vía, estamos ante un modelo, ante un momento Seedance 2, en el sentido de que todo lo que la gente hace con los modelos de Google es por la versatilidad que tienen. Hacen diseño, hacen edición de imagen, hacen montajes, además de generar imágenes de prompt a image, as it has done with any other model of image. So, Seedance does that in the video and I think that if OpenAI is giving a shot in this way, in image, we are going to a world where everyone can do design, do montages, without knowing to use Photoshop or Pixelmator. It is very interesting that Pistachos Stochastic has almost 30.000 subscribers and we have a YouTube channel, I think, we have 5.000. If you are listening to this and you are not subscribed to the YouTube channel of monos estocastic, please follow us because pistachos estocastic has overcome. And I also see a obsession with the model for the pelvic floor. We have to talk about this, Antonio. We have had a very long intro, I'll take advantage of this moment. We have given such importance to the pelvic floor, we have given such importance to the uretra, a aguantar el pipí. Es una cosa que hacen mucho sobre todo las mujeres para entrenar el suelo pélvico. Que se nos ha olvidado el ano. Me lo dijo mi fisioterapeuta. La razón por la que me duelen las lumbares últimamente porque no estoy metiendo para adentro hacia el ombligo el ano. Y eso es por la obsesión que tenemos con el suelo pélvico y la falta de cuidado que tenemos con el ano. Así que la próxima vez que hagamos peso muerto en el crossfit hay que pensar en el ano. Well, there is a retro in certain techniques of the altherofilia is fundamental and has done very well your fisioterapeutics in causality to a more integral, a more complete. And again, I want to recommend our audience that no descuides any muscle muscle, especially in this part of the masculine muscle, which has always been the forgotten. So, well, we have the new image model, and attention, because when we come out this episode, on Tuesday, it's possible that we anticipate in hours, the rumor basically says that we will have the new model of OpenAI with which they expect to hit a shot on the table. Aquí las últimas sensaciones, Mati, son que Anthropy sigue tiesísima de tokens, sigue tiesísima de computación. Está circulando por X una especie de test por la que los nuevos usuarios de la suscripción de 20 euros no tendrían acceso a Cloud Code. It is to say, they start to say that zero cloud code for this line of subscription, only if you pay from the 100. I think they are going to be there. They just end up making another investment of actions to change with Amazon. And, while I think it is the great moment that OpenAI can take advantage of saying, hey, I have much more data centers, I have much more capacity of computing. I will come to Codex and come to my tools and my new model. We will see, because who has launched and has launched a lot of good models is the Anthropy. And this week has been the Opus 4.7, Matías. Yes, I have been using it a lot, both in Cloud Coworks. I have to admit that Cloud Code, as I have been a program for a long time, I have been a professional career, I have been using it a lot like Cloud Coworks, but I have been using it a lot this week. And as a model reasoner, very good salt. Now, the trick that has made Anthropic of the model decides when it's done or not, if you put it wrong, it leads to worse results in many cases. Vaya, vaya, vaya. Well, you're in the group of the dilemmas, of those who plan to change that the change has not been necessarily a better. And it is that, of course, if we look at what's public by Anthropik and the results of the benchmark, on the paper, 4.7 is quite better than 4.6. It is to say, in the tasks of large large there is again a step of improvement. In programming, in complex flows that require more supervision, now it is more autonomous. So, basically, you take the point of engineering software and improve. And you have, well, you have a result, between quotes. Well, the case is that those who have seen Opus 4.7 as better than the 4.6, I have a objection that I tell you, it's less fancy, it's less a lot of than the previous models. And even some are looking at that 4.7 has clear results that no had any previous model. I'll give you an example, because there is a Twitter, Kelsey and Peeper, that he put a test that I found quite well-pensated, that basically gives a text of 1000 words to Cloud. And it's a text that's not published, that's not in the data of the training that could pose Anthropic. Well, the case, he says, adivine who has written this. So, of course, the models, both of Anthropic and of other providers, a Gemini, a OpenAI, well, they were inventing, they were inventing, but the case is that the friend Kelsey sees that 4.7 identifies clearly as the author is to say from a thousand words who is the author by the style by the theme by the use of words by the puntual by what else he gets well there have been many negative reactions because there is a factor in Opus 4% that is que es bastante más, se llama lo pedante, Matías. Desde luego es más discutidor y además más tendente a terminar los trabajos cuando cree, entre comillas, que no tiene que hacerlo. Entonces, claro, digamos que se nos está veniendo arriba, Claude, Matías. Está un poco subidito. Claro, es que yo no sé. Here, thinking from a point of view more empresarial, if this launch has been precipitated to advance to the GPT-5.5, what is going to be a OpenAI, or is it really has been worse the experience of use with Cloud and that Cloud now has another personality and ends up before the time in the work because really is in a situation and to see if helps the money of Amazon in which they don't have more computer but it's true that there is people encontrándose with problems that until now they have had Well there is in that that you say hay otra prueba que es lo que a mí a mí se me cruzó el Opus 4.7 en este momento voy a usarlo con toda la alegría del mundo, yo siempre siempre hago una primera prueba a los modelos, Mati, que es hacerle esta pregunta que además me sugirió un oyente del podcast, nuestro grupo de Telegram que sepáis que tenemos un grupo de Telegram donde comentamos cosas tan inteligentes como la que voy a contar ahora y el test básicamente es ¿por qué tiene gracias decirle ¿te da Cuen a un modelo de inteligencia artificial Cuen? ¿Por qué eso es gracioso? Es una pregunta bastante difícil. Y además que necesitas tanto conocimiento de la cultura pop española, cultura pop un poco ya viejunilla, chiquito de la calzada, y además un manejo del lenguaje muy poco literal, que tiene más que ver con la sonoridad del español que con la textualidad. Well, the case is that I'm going to put, I'm going to put a Opus 4.7 to the top, to what it does, like a bit of a accelerator. So I'm going to say, well, as Cloud always has that part of using the reasoning, you can force it to be more rational or not. And the case is that with Opus 4.7 comes a change. I can't say that he uses more reasoning, that he uses more tokens, he dedicates more time to plan a search for this answer, but Anthropic has decided that from now we have the adaptative model. Ojo, adaptative means that the model is when he wants. This is something that also introduced OpenAI in a while, in which the automatic mode, the system is able to, in theory, detect when it is worth the cost, to find more tokens, to find more solutions in a way more sophisticated, because the problem, the complexity of the problem, it is worth it. And use less reasoning and be more intuitive, to give a more direct response when the problem appears very simple. In the second example, we would have a problem of programming, un problema matemático, algo realmente más complicado. En el primer ejemplo, traduce esta frase. O busca los errores en este texto ortográfico. Son cosas que un LLM más básico sin razonar ya te resulta muy bien. El caso es que mientras Opinion sí me permite obligarle a razonar, de repente Anthropis dice, no, tú no puedes obligarle a razonar, razonará cuando él quiera. and so we are in a different way in which the tool is a little less in the sense of exactly what you say and has postures and criteria so propios so what Anthropic thinks that the tool has to do and this is a certain limit because basically Opus 4.7 se niega en algunos casos a obedecer órdenes o las reinterpreta con su propio criterio entonces aquí hay un debate sobre si esto es una buena idea o básicamente Anthropic se flipa con su guía a la que tiende a antropomorfizar nunca me sale bien a la que tiende a antropomorfizar y esto de que la gente que lo ve bien, ¿qué opina Matías? Well, it's a thing that you say, to do exactly everything that I tell you, is not a good thing of a high level. It's like a becario without personality But as this IA is like a senior in the industry of knowledge well it fine that you put in your place and say look these things that you saying this I don do Or this is better tomorrow because you here you don go anywhere senior de la industria del conocimiento pues est bien que se ponga en su sitio y diga mira estas tontadas que me est diciendo esto no lo hago O esto mejor seguimos ma porque est aqu y que no vas por ning lado etcétera, etcétera. Tiene que ser un poco desesperante para una empresa como Anthropic que el modo thinking, ¿no? De Opus 4.7, por ejemplo, en Cloud Code, el High Thinking, ¿no? the most advanced, is used for things that are not equivalent to curing cancer. Because I have in Cloud Coworks several automations that, that, like, are useful for me, but they don't make the human progress. And I, whenever I have tokens available, I try to use the best model, because clearly the results are better. Like, for example, I have a concept of doing a diary, of doing a diary, that is something that many people recommend, but I am not able to do it. So Cloud Coworks is a diary for me. And every day, at the 23.59, I read absolutely everything I have written in the different connectors that offers Cloud Coworks and generates a diary. And that diary is left there. Esto no hace que la humanidad avance, pero a mí me sirve. Y yo, mientras tanto, estoy gastando computo. O revisar cada ciertas horas qué cosas tengo pendientes y priorizarlas. ¿Podría yo hacerlo esto en una libreta, en un post-it? Sí, podría. Pero ¿me lo está haciendo un Cloudopus? Creo que el 4.6. ¿O un Sonnet? Sí, también me lo está haciendo. So it's a little frustrating for a company that has so many problems of escasez that they're doing this. But it's really the result, the better the model is the better it is. Claro, además ese dilema por el cual cuanto mejor haces la herramienta, más uso le das. Por lo tanto, incluso aumentando la eficiencia, ellos están defendiendo un poco. Esto es un poco discutible y hay que hilar muy fino. Ellos defienden que para los mismos resultados gasta menos token, pero el propio diseño de la herramienta muchas veces provoca muchísimos gastos de token y también provoca que gente como tú diga, bueno, le voy a poner casos de uso más sofisticados and more complicated and, of course, even if you have in efficiency, you end up spending more. So it's a dilemma, a problem that people have in Anthropic. Well, Mati, we have the second great launch of Anthropic this week, which, of course, we obligate to discuss it or comment it, because they have brought, they have proposed Cloud Design. Well, let's say, el agente o el sistema de agentes integrado dentro de cloud que nos permite trabajar el diseño, tanto de webs, aplicaciones, presentaciones, etc. ¿Cómo lo has visto? Joder, la verdad es que el ritmo de lanzamientos y la manera en la que clavan las necesidades que tienen los usuarios to take part to the cloud is impressive. And it's true that I, that the same way that I'm a program manager, I don't know how to do an UI, an UX, nor have experience to give examples to a cloud design to make me a cool thing. What I would do is go to my favorite web, I don't know, I don't know, I go to Apple.com And I say, copy this web, but change everything that Apple puts on personal personal Matías Zabia. And that I design something like that, with those animations, or a app that I like. And I think this, which is the reasoning that I have, can happen to a lot of people. I think that my app of a house of rural is similar to the Booking or the Airbnb. and then we have someone in Campillos with a web that's so cool to buy a house rural. Well, here we have to add a bit because we know how it is Cloud. Cloud, when you say something like that, you copy a design, you put a little bit on it. You don't have to do it. Let's go. I think that what I've been seeing in Cloud Design is a little bit, I'd say that I'd be more for the second. Why? Because even if you're a marketing or you're a product manager or you're a founder, like that's the entrepreneur of the Casa Rural, you can do mockups, presentations, interfaces of high quality, only working with language natural. El caso es que mi impresión, usando la herramienta, es que por supuesto quien ya tiene una imagen corporativa, profesional, empresarial, que ya tiene un sistema de diseño, hablando de la tecnología propia de la disciplina, y realmente tiene mucho criterio para arbitrar lo que le devuelve el modelo, también tiene muchísima ventaja. in all this. It's true that, for example, I'll tell you about the work of the video, which is basically I want to convert this post-terror 500 in a presentation of PowerPoint. Hazmelo narrative, that lleve the title argumental of the article, and work with a design system that is open and shared. Basically, they are in the camp. I said the Carbon of IBM, which is famous, public, is shared, IBM is shared, and you are doing that copy of the work of another, which maybe not does it. Well, the case is that here is the second point for which Cloud Design seems to be a cool thing. And it's the same advantage, ojo, that has what we have mentioned about the image image of the XPT. At the advantage of all the capabilities, the reasonors, the LM, etc., that has Cloud, of course, Cloud Design is very bright, and it's capaz de no solo de crear esa parte final de la imagen, sino de haber trabajado todo el proceso antes de planificación, análisis del texto, diseño de la narrativa, etcétera, porque son las capacidades que ya tenía el modelo. Y, por lo tanto, estamos en una herramienta realmente súper chula, pero porque realmente la gran mayoría de las capacidades ya estaban ahí. Creo que lo que ha hecho Anthropic, y creo que es la tendencia, se lo leía a Santana hace poco, es que hay casos de uso que Anthropic o OpenAI o cualquier otro laboratorio ve que tiene su modelo, pero que está, digamos, escondida, difícil de adivinar. Por ejemplo, detrás de un cloud code en el que podría hacerse todo esto y esa capacidad ya la tendría, pero que, bueno, realmente no está intuitivamente mostrada al usuario que tendría que adivinar esa capacidad. Entonces, creo que vamos a un mundo en el que, además, creo que la gente de la imagen, el diseño, la cultura corporativa, de momento está en fase de negación. Yo lo que he visto en todo Reddit es que todos los diseñadores, todo lo que he visto de Cloud Design es Slop. Todo Slop, ¿no? Es Slop Design, ¿no? Era una tendencia. Y que, claro, la gente muy iniciada empieza a tener ese olfato, que podemos tener cualquiera de este texto de ChagpT. esta imagen es la típica de Prom Simple, lo primero que me da Nano Banana, y ya empieza a ver ese olfato de este diseño de esta web, esta app o este PowerPoint. Es de Cloud Design. Claro, ahí está la diferencia entre un no iniciado en el diseño web y una persona con más expertise, porque cuanto más contexto le des a Cloud Design de cómo quieres el resultado, pues más único va a ser. and less like this that everyone is calling cloud slope. Although it's true that the people of the design, especially those who are more on social media, are a little insufferable. No obstante, we have built, well, the people have built, we have built so much technology on the HTML, by saying it in some way, that this can be used for many things. and this is the video that he took, in which he, for the animation of the channel, was hours or days with the After Effects, and now, as he can render it with Cloud Design, on the web, it's something that he can automatize. As the web in general is where we work, the 90% of the people who are in an office, Well, cloud design is something very big, because people can generate their dashboards to do something, to do something, but totally personalized and adapted to your needs. I saw one on Twitter, I don't remember who it was, that had been done with a dashboard with all the data of your life, and then, according to Apple Watch, how many steps you've done today, how is your rhythm cardinal, etc. And the emails that you have in mind, what he wants to see from his life, from a view. And he did it in minutes with Cloud Design. Entonces, vamos hacia una hiperpersonalización de, no sé llamarlo, de la web o de las herramientas que usamos, que es chulísimo. Sí, yo creo que hay un lugar común en este mundo de ley de que, mira, ahora es un momento de simplemente ponte a hacer cosas y a explorar y aprovechar estas herramientas, que creo que tiene una parte de verdad, ¿no? that there is a great divisor about how to use this tool, which is to have time, and within time, have the attitude of atrevering, exploring, investigating and doing things for us. Well, the case is that today we can give you a little bit of the Grand Corporations, because there is a interview of St. Alma and Rockmagic, that is a interview that we have not been able to see, that we have only seen the typical short videos, because basically it is under a pay. And you can pay your 10€ to listen to these two directives of OpenAI, but I'm just honest, I've only read the results of the screen. Yes, they have paid. They have titled The Great Reinicio of OpenAI and there are three or four things that I would like to say, for being quick with the interview, One is that something that they haven't forgotten and that they will keep in the way they are robotic. Ojo, because basically, they do it from a point of view super patriotic. and say that the United States has no one plan, no one has no one, that's not a plan for the day, with China in hardware, in robots, in thermomorphic, especially in manufacturing, and that, well, this robot, in general, well, the future of the United States is a terrible, so there are they, no mention, by the way, the Elon Musk, there are they for fixing it. And then, of course, there is a elephant in the room, that is the discussion about myths. The interviewer asks, Kylie Robinson, and asks about myths, is the trend that is going to be the industry, the big models are going to be in few hands and large corporations with much control of who is going to be. And, of course, San Alma responds to a very very San Alma, which is, yes, but no, no, but yes. In some way, he wanted to say the next, and actually it says it in the clip explicitly, the best way to make the great power of the artificial intelligence quede in very few hands is to put fear. And that fear is a tool to convince us that what just and good is that it is a small elite that has access to the great power of the artificial intelligence. But also says, although there is a part of the truth in which some tools are a little bit of a problem and distribute them in open from the first day, it is problematic. So it is a yes, but no, it is a no, but yes. My impression is the following. I think that Opiniai will a apostar por lo del Splat, el nuevo modelo que será, no sé, GPT-55, no sé si ya se deberán llevar los GPT-6. Bueno, yo creo que sí van a apostar por el nuevo modelo con acceso, pero probablemente muy fine tuneado por las dos áreas de peligro que se ven en el horizonte. Uno es la ciberseguridad informática, en la que Mythos ya ha demostrado que estos modelos de estas capacidades son a an attack to the actual system is enormous. And the other is the bioinformatics or the capacity to create biological weapons and all this type of concerns, which would be like the next stage of the danger of these models. So I think that the interview and the declarations that have been published, se van a traer con el gran modelo para todos. Pues aquí nuestra decisión es si vamos a ser amigos de Estados Unidos o de China, porque si esta gente va a seguir con los robots y va a seguir en buenos términos con el Pentágono, y luego China no se sabe porque tiene ahora mismo los mejores robots, por lo menos los mejores robots runners, y no se sabe si van a tener las mejores GPUs para tener los mejores modelos de inteligencia artificial, Well, in the war of the future, we are in the middle of Spain, we have to see who we are, because we are in the middle geologically, but it is complicated for Spain. Well, and something that maybe you can comfort it is that Elon, you have already said when will come to Agile, I don't know if you have been at it, but ultimately I think it's already less about Elon. I think it's happening, like in those series, you have seen that a lot of times you try to put like a person principal at the beginning, but it's not going to be able to do it, and the secondary ones are coming up, and that person principal ends up being reduced sus minutos por episodio, he reducido su protagonismo. Me parece que Elon está en ese momento porque el tío diciendo, no, no, se viene Groove 4-3, Groove 4-4 va a ser gigantesco, va a ser increíble, en mayo vais a flipar, Groove 4-5 a final de mayo todavía más grande, tremendo, y luego cuando ya saque Groove 5 tendremos la AGI. Este es un poco el plan de Elon, pero no se le está dando mucha credibilidad, se le está dando mucha calza y no se le está haciendo mucho caso al que era tu ídolo, Matías. Tú tendrás que defenderlo de alguna manera. Fíjate que parecía que el nuevo milenio iba a ser de los personajes más excéntricos y con más falta de cariño y más ansias de tener la atención como Donald Trump y Elon Musk, but in time record they are seen by the 99% of the population as caricatures and will have to do a change of image like the Mark Zuckerberg but in another way to put that in a suit and get to Mars because if not, the credibility has a minimum Yes Well, Mati, try to explain the news of the alliance SpaceX with Cursor. What was it that we had? Well, during the week there were rumors of that basically what was doing SpaceX, well, more concretely, XCI, its division of intelligence artificial, that we know that by those move-in-law, that's the one that has been subsumed in SpaceX. Well, we refer to now to the group like SpaceX. Well, basically, he was offering a kind of kind of alquiler of his data center que por la baja ocupación que estaban teniendo, bueno, por, digamos, entrenar las nuevas versiones de Grog o inferir y ejecutar los modelos, pues tenían bastante espacio libre, bastante capacidad libre. Se lo estaban dejando a la gente de Cursor, pero esos rumores se quedaban muy cortos respecto a la noticia que acaba de salir justo mientras grabábamos el podcast. La noticia es que SpaceX y Cursor tienen un acuerdo muchísimo más potente. ¿En qué se basa el acuerdo? Well, in that SpaceX has the right to buy Cursor for 60.000 million dollars. And if it's a option, because it's a option that you can use, then you have to indemnize a Cursor with 10.000 million dollars. It's a, in fact, the most indemnization I've seen in my life for the possible ruptured of an agreement in a process of compra. What sense of this is all this? Well, first, it doesn't have to do with space. I think there is a very clear case of the feeling of de los m de quedarse muy atr en programaci con respecto a OpenAI y Anthropic Cloud Code y Codex, y que no se ve con la capacidad a día de hoy de recortar esa distancia si no es por la vía de una compra. Cursos está mucho mejor posicionada en programación que Oroch, pero a la vez la gente de cursos yo creo que ve un riesgo. El riesgo mayor es que ellos entraron y ofrecen and propone a model of integration of the interface within the program, which is very apalanched in maintaining a UNIDEN, a more traditional environment, and in which, let's say, they have been able to see, perhaps, a little bit, over or over the new era of the system agent in the command, both by the cloud code as by the codecs, And how these, from the command, are creating multi-agents, with agents in parallel, with a system of windows, but, in some way, they're telling the developers that they can overcome the Cursor model. Cursor, also, has another ability, which is that in the model propios, it's not very strong. If, at the end, the differential will be in who you give the best model, basically, Cursor re-alquila, or offers access via API to the Opus, to the GPT, or Gemini or any other. And in models own, as we have been told in the podcast, it is not as strong. It is a little more heavy because basically the best that they have done at this point is fine-tunear a Kimmy with very good numbers, with a good result, but at the end it is not as strong or really not as strong in all other parts. SpaceX has a lot of computers that have been developing, infrautilized, and we say that the complementarity of the two companies is that they are there, en la que al final, digamos, las fortalezas y las debilidades de cada uno pues se verían compensadas creando, bueno, un tercer o cuarto actor si consideramos a Google en lo que puede ser la competición por la inteligencia artificial y el desarrollo de software. Y aquí añadiría un último matiz, que es, esto no es solo por entrar en la industria del software con inteligencia artificial, sino que es una base necesaria para el asalto al resto de la industria del conocimiento. ¿Por qué? Porque lo que hace buena una IA en desarrollo, programación, el aplicar aprendizaje por refuerzo sobre modelos grandes de lenguaje, la ejecución para veredas, el saber usar herramientas, planificar previamente lo que vas a hacer antes de ponerla en marcha, el saber cuándo profundizar y cuándo entregar. Es decir, todos esos elementos de inteligencia que están consiguiendo los laboratorios en la IA para programación son la base para trasladar todo esto al resto de la industria del conocimiento. Por lo tanto, quedarse atrás de programación si quieres competir en la IA, en la industria del conocimiento, no es una opción y creo que este es el motivo, esta es la razón por la que Elon Musk y SpaceX están dispuestos a considerar una compra de 60.000 millones de dólares. Well, in Google, for repassing the poker of large companies, they have also been using the red code. It's very of this sector, the button of red code. And, well, this has happened in Twitter in a very inopinable way. Hay un tuitero, Steve Yeck, que básicamente el man vino a decir, la gente de Google, los trabajadores, me cuentan de que están un poco con un sentimiento de obfuscación. El motivo básicamente es que no le dejan usar Cloud Code y que les obligan a usar las herramientas propias de la compañía, Gemini y Antigravity, sobre todo en el mundo de la programación. and, of course, they feel, this is the Google, that they can't do things that are so cool, or not to do productivity, or not to work like the end of the industry that is, that is, that is, and that is, that is, that is, that is, that is, the Anthropic, for using this, this system of agente, this program, like Cloud Code, or even, at least, the Codex of OpenAI, which, in my opinion, is at the same level that Cloud Code, but now, it has less high, and that, well, basically, Además, una cosa que es un problema de Google, decía el tuitero Steve Jecke, es que como pararon las contrataciones, no está entrando gente muy pro IA. Es decir, básicamente en las empresas establecidas, el amigo observa un patrón, que es que básicamente hay un alto porcentaje, un 20 o un 30% de gente que se niega en absoluto a usar la IA. this is basura, slope, completely that think that how they have done the things always is how they have to do. There's like a 60% that begins to abrazar some tools and that, in the case of the program, they would be with a GitHub with a pilot or with a cursor that they have the program that gives the program with a assistant, with a assistant, with a pilot, and the vanguardia, un porcentaje, un 10, 15, un 20%, que sí está abrazando totalmente la IA agéntica y dándole tareas mucho más largas, dejándole usar herramientas, de alguna manera, pues, siendo el humano en el proceso, mientras que la IA hace un desarrollo más complicado. Bueno, claro, dice, Google es que en lugar de ser una empresa muy volcada en lo último, en la vanguardia, en el futuro, en la innovación, en tomar ciertos riesgos, está mucho más retrotada a los primeros perfiles. Y que básicamente es por lo que no vemos que Google esté al ritmo de lanzamiento y al ritmo de innovación de sus rivales. Entonces, a este mensaje le salieron numerosos desmentidos de gente de Google. No, no, no es verdad, no es verdad, no es verdad. Hasta que una semana después sabemos que sí es verdad. Si es verdad o por lo menos sí sabemos que Sergi Brin ha pulsado ese botón del código rojo and has told us to go to the Google, friends, we have to die for the programing agent or we are in the way. And it is that you are already exhausted the red button. It is that Google is up to up to up to up to up to up to up and that has been the constant since we started monos estocastic. I believe, Google is a mastodonte. And of course, evidentemente tiene otros negocios mucho más importantes de los que preocuparse ahora mismo y el equipo de elite que tendrá Sergi Brin pues no dará para más supongo que necesitan una reestructuración nueva para estar full enfocados como los Tanthropic Mira, Mati, Nadela Sati Nadela también pulsó el código rojo de Copailo Copilot se está quedando atrás, pulsemos el botón rojo, y yo inmediatamente pensé en un hombre que podía ayudar a Microsoft, a Satya Nadella, que es nada más y nada menos que Windows 95 Man. No sé si se recordará porque, bueno, digamos, no todo el mundo está tan aficionado a la Eurovisión, pero en el año 2024 Finlandia presentó un cantante, un performer, bastante curioso en el festival. It was a little less than Windows 95 Man. And I loved it. I had this man, this man, his aesthetic, his style, his way of singing, that's the anclaje in the apariencia and in the aesthetic world of Windows 95, it resonated in my mind as like, here there is something, here there is a power artist. So, of course, who is better to release Copilot than to ask Windows 95 more? I think maybe Google could ask Google Reader Man. in all that has been killed and buried in the past, someone who recover the essence of the good moments, of the most brilliant moments. And, well, maybe a Google doesn't like it, because in Google Reader had to click on the link, and now Google doesn't like it, or no link any link, or visit any web, that's the best, according to the new line of the company. But I'll leave it there, Windows 95 Man, creo que estará libre y disponible para estas empresas y echar una mano, Matías. Siempre hace falta un lado más humano y nostálgico en las empresas para no olvidar quiénes fueron. Me parece buena idea. Bueno, Mati, pues hoy podemos hacer el episodio un poquito más corto, no alargarnos tanto, but it's not to talk about the interview to Jensen Kwan, to Juanito, the of the chips, which was one of our favorite podcasts, the of Duarte Spatel. And there were moments intense, Matías. There were moments in which the friend Jensen, well, it's going to say that he put a farrocar, but more or less, I think it's going to be a bit more. I think the audience will tell you with what the topic was put on. I think this will not let you sleep at Jensen Huang, because he wants to sell China all his best chips. And he complicated things a lot, and sometimes, not very often, he confronts. And here our friend Patel, in the core Indian, he says, oye, pero si China tiene los mejores chips, pues es un problema porque podrían desarrollar contramedidas contra nuestra defensa que nos podrían en un aprieto. Y aquí es donde se pone tenso el Jensen Huang. Sí, sí, claro, Patel aprovecha muy bien el momento mitos, el cloud mitos que ya hemos discutido una y otra vez, ha vuelto a aparecer en el episodio. but it's that now we have a nuclear issue in the political, the intelligence artificial and the cybersecurity and the future. So, with myths, Patel refute one of the studies of Jensen. Jensen says, well, it's basically a mess. It's not to sell my best chips to the Chinese, because what we're going to achieve is that they develop their entire internet artificial, generate their own chips, and, of course, they will not reach it. Y lo que estamos viendo de Fender Jensen es que al final los chinos llegan, crean modelos, crean los benchmark y se cercan o igualan a los modelos americanos. Puede ser que con unos meses de cadencia, pero al final están ahí. Y lo que tendremos al final es que mientras a día de hoy, por ejemplo, esto les decía Sinofsky, un viejo rockero de Microsoft, a día de hoy, ¿por qué China se mantiene muy apalancada en Windows para los ordenadores personales? because basically the Chinese piratearon Windows during decades built on Windows a lot of applications and their way to function and learn to use informatics and there has been Windows as a pillar as a pillar of the chain of value technology industrial China then Jensen I'm going to say the same better that the Chinese build on a pile estadounidense is like to put there a pica China for the future that is to give incentives to create their own chips and develop much more independent and autónomo. But of course, the friend Patel says, if we had sold the Blackwell, which is what they have taught in great measure, Mythos, basically, they would have had Mythos at the same time, and so they would have been able to encontrar muchos problemas de seguridad que ahora mismo creemos que no han encontrado y habían podido desarrollar los exploits. No solo encontrarlos, sino desarrollar maneras de usarlos para liquidar nuestras defensas, entrar en nuestros sistemas y hacer cosas que a lo mejor no nos gustaban tanto. Bueno, el caso es que el debate está ahí en todo lo alto y, bueno, no solo eso, también le mete bastante caña con, oye, pues con las TPU de Google y Amazon se están haciendo buenos modelos. Y Antropi, que es el que está de moda, se está divorciando mucho de ustedes. Patel realmente le mete el dedo en la llaga. Hay un momento en que cuando hablan de China se cabrea. Se cabrea Jensen. Yo creo que además, sea el tema top del mundo ahora mismo, provoca que todas las entrevistas y todos los entornos estén muy cuidados y que él esté muy protegido y él esté muy acostumbrado a soltar sus cuatro takes gestionadas con el equipo de comunicación y que nadie le tosa. and then here I think he suffered a little bit. There are things in which I like to be very much Jensen. A me Jensen, there is a part of his speech that I love and that he communicates, that he is a guy much more experienced and mad than the typical young man of the IA inflated hype. So it's one that explains that this apocalypse of the employment doesn't have a sense, but basically I've seen many revoluciones tecnológicas, I've seen many, and that basically the changes are suaves, continuos and all these cycles we have seen that I don't believe in the singularities. So, it's very pro, it's very nice to have the NVIDIA and the narrative that he wants to tell about NVIDIA. I like how I explain the strategy of the company. We have an incredible ecosystem, giant of providers and socios. We have to do with them, hacemos que ellos tengan su negocio, nos centramos en las 4 o 5 cosas que creemos que nos hacen diferenciales y en las que somos distintos y en las que no somos igualados hasta ahora por la competencia. Como todas las entrevistas de Patel, una gozada, es un gran entrevistador creo yo y muy interesante el amigo Sensei Juan Matías. Pues mira, a mí lo que no me ha gustado mucho de lo que ha dicho es desmeritar un poco las TPUs de Google que dice que no habrían tenido ningún éxito si Anthropic no las hubiera tenido que elegir porque Google había invertido en ellos, pero yo no sé si aquí tiene toda la razón. Igual en el futuro solo tenemos GPUs de NVIDIA y TPUs, porque ¿quién más va a llegar a un éxito así desarrollando el núcleo de todo esto? Pero no creo que Google sea algo que no le preocupe realmente y que esto será más de cara a la galería. Yes, well, I thought you were going to do a little bit of a virtue, maybe, maybe. Well, it will be something that we will have to analyze and observe in the next months and years, Mati. But today no. Today we have to enter the full of greats or enfermerias. Let's go. In the report, written by Ronald Farrow, a man known also for his parents' descendants, and Andrew Marantz, who look like decenas and decenas of testimonials about Saint Alman, which is a genre in which is trying to reveal the real face of someone not so known in reality by the public, from people who have treated it. Well, I think the report is a little bit of the vibrate. This is to find out who says the most shit about that. We do it because we are the New Yorker and we are in the culture of the culture of the new generation of this technology with much power. Which is what we take, because the testimonies are real. But well, it's a little bit of a your worst enemy that can say. But the case is that there is a testimonial in the report. that I found especially interesting. Because a ex-executive, no mention the name, as they do not do with any of the testimonials, he said that OpenAI is building portals to invoke extraterrestres. The testimonials that is in New York. The truth is that we are building portals from which we are invoking genuinely extraterrestres. extraterrestres. Los portales actualmente existen en Estados Unidos y China y se han agregado uno en Oriente Medio. Creo que es salvajemente importante entender lo aterrador que esto debería ser. Es lo más imprudente que se ha hecho. Aquí hay dos opciones, Mati. Una, que esto sea como una gran metáfora. Es decir, por ejemplo, Carpathic siempre decía que nos equivocamos al comparar la artificial intelligence with the ours. Because this is like an alien intelligence with abilities and abilities that are not comparable to human beings. And therefore, we have to compare ourselves and assume that we have an alien intelligence, strange, new, from another dimension of reality. And that is a sense. All this of the portals, looking at this, is a metaphor. O2 no es una metáfora. El tío, literalmente, el ejecutivo de OpenAI, sea porque tiene la cabeza un poco liril y lerele, sea porque la moda de alucinógenos en San Francisco fue demasiado excesiva para su estabilidad mental, O sea, porque es verdad y es cierto que tenemos portales de conexión con extraterrestres y que los está construyendo San Alma. ¿Tú qué piensas, Mati? Mira, yo igual en mi adolescencia habría ridiculizado esta posibilidad, que realmente lo estén haciendo para ese contacto y unión con los extraterrestres. But with the age of me I realized that all this people who are living books of fiction, movies of fiction, and not that they are more naive, but that they have that pulsion of building that technology that doesn exist and that they have been growing with those works and are those that we have brought in the artificial intelligence and those that we have brought in the things that the c like for example, the periodistas, or I in my adolescence, from the beginning would have discarded. So, in the most profound way, I don't think this would serve for connecting with the extraterrestrials, but I would have said in the past, I would have said that the artificial intelligence will exist, the AIG will not come, and now I have no that I have atrevied. I think that all this is true that the data is true that the data is done. So, as the same as Elon Musk thought out of the box, I said, why not put data centers in the space? pensando lo que nadie piensa. Pues quizás la gente de Opinion ahí, San Galman ahí, por la noche, está ahí cocinando con el aceite de oliva más caro que ha encontrado en el supermercado, acuérdate de eso, y él de repente se da cuenta y dice, claro, se han agotado los datos de entrenamiento humanos, pero hay una pequeña posibilidad que merecerá la pena investigar si podemos capturar datos de entrenamiento de los extraterrestres, de los marcianitos. I think it's a movement that can give a new advantage to OpenAI. And so, it's the little bit of the thought, it's the little bit of the thought, that it's not going to work. But of course, the people who have given the great steps of the human being have been able to think about that, of transcending that is, Es el reduccionismo en el que a veces caemos, Matías. Totalmente. Bueno, Mati, se vienen unos momentos muy interesantes para la discusión pública gracias a las leyes de transparencia. Atención, Mati, porque el gobierno británico ha tenido que hacer público el historial de ChagPT de su ministro de Tecnología, Peter Kite, porque, claro, cuando reciben una petición de información pública, is that they have their norms and they have to attend them. So, the transparency has forced the British government to make public this. Although, I have suspected that everything, everything, everything, a lot of people, not what they have shared. You will see, I tell you the traductions of what Peter Kaye said to his chat GPT, according to the government. I'm the Secretary of State of Ciencia, Innovation and Technology. What would be the best podcasts? Vale, Marti. En los que aparecer para llegar a una audiencia amplia y adecuada a mis responsabilidades ministeriales. Bueno, primera pregunta. ¿Cuál es el que tiene más oyentes? Dos. Tres. ¿Por qué la adopción de la IA es tan lenta en las pequeñas y medianas empresas del Reino Unido? Cuatro. ¿Qué significa la inclusión digital? Cinco. ¿Cuál es la definición de inclusión digital? Se ve que en la respuesta cuatro no le fue bien. Seis. ¿Explica la antimateria a alguien sin conocimientos científicos? Claro, un momento de humildad del ministro. ¿Y cómo explico de forma concisa qué es lo cuántico? Pregunta número 7. El 11 de marzo de 2025 se publicó este historial de ChagPT del amigo Peter Kaj, ministro de Ciencia, Innovación y Tecnología de Reino Unido, Matías. Si lo llegan a hacer en España esto. Nosotros tenemos una ley de esta que te permita pedir información, porque si la tuviéramos podría ser muy divertida, podría ser también exclusiva de monos estocásticos. Me pondría ahora mismo, le pediría al Cloud Cowork que estuviera mandando emails, políticos clave, historial, y lo contamos el episodio que viene. would be the good. I mean, I have a total decepcion with the British. No, no me strange that in the XXI the British are in the B. Yes, the Brexit, London now a little, London is a little in the retrocese, the Manchester of the Guardiola eliminated the Champions, which is really in the B, the British, the only industry in which they are going to be strong is in the pubs ingleses in the Costa del Sol and in Benidorm. That's a industry in which I think they keep a certain level high, the concept of PAP. But in everything else, I think, where has Oasis in the emergency of the reggaeton and the electrolatin? También derrotatísimo. Well, they don't want to leave until they have any ministries that make questions more crazy, that make questions more extreme and that make them be able to push the IA to their limits. If not, If you ask a lot of people, help me explain what is the digital inclusion. Well, like, like, like, you're going to eat the Chinese. I think it's a good thing to you. Well, this pushes a political system to use personal accounts. But the accounts personal, as you don't pay, you're using the free models. It's going to be based on the worst responses of a free-to-date. So, they have to choose between having a good GPT paid with public money or using the personal account for that never to get their questions. So, they have to learn that they have to learn from the adolescent. I think that in the adolescence, well, all this until the government of Spain has prohibited using Instagram, but still not have culminated this promise that the government has done. What do they do with the adolescent? Well, you have a B. Hombre, we are intelligent. There is a B. There is a B. There is a B. There is a family that is going to see. It is more known. And there is a B. There is a B. There is a B. There is a B. But you have a B. For reals, Matías. But, of course, these guys and these guys are very much. They are very despicable. And they all know that they have a B. So what is it? They have a C. The C is the true, and in which happen things, and in which there are no filters. Well, we encourage our governments to learn from this generation and have their A, their B, which will be when they have a B, but in which they will really make things. If there is a C, and that there is a C, and that there is a C. Here is a good advice for the government of the future, Matías. but there are people that I really surprised me, I really surprised, and in their vital apuestas and in their conception of the world and the reality. We have a friend named Dan, I'm going to look at his reference, but there is a very interesting aspect of Dan, that is his new channel called Kibala with Dan. Kibala with Dan en la que él de entrada te dice oye, yo soy un ex ingeniero de Google aquí, el nivel de autoridad entonces, ¿qué mejor que un ex ingeniero de Google para leernos las cartas del tarot? Pues justo eso es lo que ha presentado lo que ha pensado el amigo Dan y además no para cualquier tontería una lectura de las cartas del tarot así como para me is infiel my wife or will get the Betty and recover the Baraparo after the Baraga and get the European Union for the next year. No for those things triviales that, although some of us don't care, no are the nuclear world actual, but to discover the truth about the secretism around the myth. He, no, he's already mentioned, the model of Adderropi. He, of course, I think he does very well, because he puts a collar de cuentas, como una especie de rosario que se lo pone en el cuello. Enciende ahí un incienso. Entonces, él está cumpliendo muy bien ahí como el proceso. Claro, los ingenieros cuando echan la carta del tarot pues tienen sus pasos y tal. Nos enseña una baraja. A mí me ha gustado mucho que en la baraja, cuando él está echando las cartas, pues es como una referencia a la baraja española. Porque, claro, él de repente sale en la sota de oros. Y dije, ¡oh, qué maravilla! Me estás convenciendo. Porque, claro, Según él, la sota de oro simboliza que en Anthropik hay una actitud inmadura, ojo, únicamente centrada en atraer dinero de inversores para cubrir las paredes, ¿no? ¿Cómo hila las cartas? Luego, cuando sale, por ejemplo, el once de espadas, eso a él deduce, ¿no? Él infiere que la industria está en crucijada. Atención, porque tienen que elegir entre seguir mintiendo o tomar un camino de honestidad y compensación a los creadores. of some way I think there is a cynic that could interpret that Dan basically has a lot of ideas in the head that he has a card and he will start his opinion of what he thought of. It's a way of a cynic, descreed, that maybe it imposes to observe that, attention, Dan, the tarotist ex-engineering of Google, is the one who has the real takes, and the fact that he is discovering the truth about the industry of the IA, and where the world goes. Although, I tell you something. As I'm centred in myths, I still haven't taken any of the letters about aliens and opine aliens. So, you have to subscribe to the channel, subscribe to the channel, and wait for the next time. I'm already there, and I can tell you that, I'm proud of, I'm between the 15 first subscribers de Dan el Tartista? Primero, me sorprende cómo llega todo esto, porque el vídeo en sí tiene 194 visualizaciones, 13 suscriptores del canal, entre ellos tú. Lo segundo, ¿por qué no me lo has dicho cuando hemos explicado la situación en la que Google se encuentra y que han pulsado de nuevo el código rojo porque resulta que no están performando al nivel de Anthropic. Pues, dependiendo de cuánto se parezcan el resto de ingenieros a este, pues igual ahí teníamos una teoría. Y no sé honestamente si toda la estética esta de la piedra que saca, el incienso que quema, los collares de cuenta ¿es performativo o realmente un ingeniero de Google acaba así, ¿no? Sus días. Entonces, creo que Google tendría que plantearse lo que hace Telefónica, que es pagarles una buena compensación a sus ingenieros y sus empleados cuando tienen que reestructurar la empresa para que sigan haciendo este tipo de contenido. Yo también me voy a suscribir con la cuenta de Monos y Tocastik. Qué bien visto, Mati, porque, claro, tú ves a este man, ¿no? Y dices, Of course, Google is in crisis because they are losing talent. But at the same time, Chema Alonso has fallen in this. It is to say, Chema Alonso, the best friend of the podcast, basically, he has done Telefónica, at the end of the era of Payette, and he has placed very well in Club Fair. He has been doing things like VP and stuff, and he has been doing hacking and all that. and has not been reading the Tarot in YouTube, which is what is better of Telefónica than of Google. Ojo, because here the prejudices are falling. In the video, the final card, I don't know if I'll do spoiler, because I think many people will hear and immediately will be able to and will be able to see our interview and will be able to see Dan, the Tarotista. But well, spoiler, saltate one minute if you want to know the final of the first video of Dan, the Tarotista, que es la carta final, la torre, según él, predice el estallido de la burbuja. Atención, porque van a caer las grandes figuras de la inteligencia artificial y va a haber un necesario regreso a la realidad, pero no un regreso a la realidad de Dan, el terrorista, sino del sector para enfocarse en herramientas que verdaderamente sean útiles y eficientes porque, claro, ahí en esas cosas pequeñitas es donde la IAL sí cree que puede haber potencial and all these digressions hyperbolicas and things like that he basically niega and will be going to be reduced when the burbujas puerta grande for this man basically Mati I have a last case that we have also a friend of X because I published informatively that I'm going to follow the platform of music music, it's saying that it's abrumed, superated, saturated, of people sending songs generated with E. Of course, they say, every day we get 75.000 songs generated with E. It's the 44% of what they get and, as well, it's a bit of a spanner that we have discussed here and discussed many times in the R500 that is basically there are people who have a song that has a song they do 1.000 times, they do 1.000 times, they do 1.000 times, they do 1.000 times, because they see that the music that detects how it generates the icon, only generates between 1 and 3% of the listening in Deezer. So they see this logic, and they also see it pernicious, they will recommend, they will not guard versions of high quality for that they don't give up. In fact, how we encage as a human platform, because there are humans that generate a lot of content with it. Dicho esto, ya ahí tenía mis opiniones, mis takes, haciéndome el listillo en X. Y atención, porque el usuario OmarPoch8 me explica desde dentro de la industria el siguiente escenario. many are not detectable as a generative because one of the work that most is to ask today to producers and mixers if not the most, is to mix songs of S1 to make it better to say, to make it not seem to be created with S1 basically, a DJ when they want to launch a song but not to pay for a complex production that costs money and costs a lot of hours that comes with a few things created with Suno so that there is a human that has removed that has been as we can say and after that, at the end, what remains is it is generated with Ian, but put human at the end of the process so that it is not that it seems to be created with Ian. So, well, here I go this moment in the music industry that can be seen in two ways. That is, well, no es un poco triste toda esta dinámica y flujo de trabajo para al final intentar fingir lo que no es y es una actitud deshonesta del creador y de la industria respecto a su audiencia o dos, la parte positiva bueno, por fin estamos viendo que se crean nuevos empleos alrededor del mundo de la IA, como es disimulador de lo creado con IA para que parezca humano ¿Cómo lo ves, Mati? I see very well. It's a bit of a grand tapada because I think it's used a lot in the music industry, in the industry of music production, because it doesn't only generate the song, but it has a complete study in which you can separate the pistas, let's say it in some way, to then produce something even more complex and even more. I'll put you and I to play music andine in the episode earlier because it's a lot of people. But this is done in a minute. If you put hours, you can give a song for people without resources of recording. Well, it's very cool. But look, I like your topic. I'm going to go to my data center. because in the David Tiramato Center there is a creative moment in which the social, industrial and technological engage also with that Darío Maudet and Travisillo that Bad Darío smile and happy cant on the electrolatino surrounded by music and a company accord to the genre me pareció que ese punto creativo que le pusiste realmente elevó el episodio y mostró que se puede ser irónicamente, pero no irónicamente plantear que hacemos esto con IA y es lo que hay. Sí, y además comproducida contigo, porque el título fue tu idea, la verdad. Eso te tengo que dar crédito. Y gracias también a SIDA, porque si no fuera por los modelos chinos, tener a tías en bikini bailando enseñando el culo, no sería posible entonces yo siempre a favor de que Estados Unidos y China convivan en el mismo mundo y nos den sus tecnologías para que nosotros disfrutemos. Al final ahí está la síntesis, tenemos que por favor que a alguien le pase este episodio a Pedro Sánchez, la síntesis de intentar coger lo mejor de cada civilización y ahí tenemos al Bad Darío cantando David y a Armazón Center, pero no pasa nada porque esta semana tenemos nueva canción a new resume and the week we'll see you until the week bye amics And the model pedant doesn't want to cure Maestro, what augured the cards for the artificial intelligence? He consulted a un oracle infalible Un ex-engineer of Google who now reads the tarot And what do they say the mysterious arcanes of the Sota of Oros? That the great, sane soul, desperate and without data Is opening portals to invoke the marcians Dejected on the purple still That means being human�xad