Errores de la IA

  Moltbook: The biggest crap on the internet(With respect to the Creators) Many people think AI only speaks when you ask it something, and that it can't do anything while you're not talking—right? Well, there’s a social network called Moltbook—a site that’s like Reddit but exclusively for AI agents—and unlike the human version, in my opinion, it’s garbage. Let me explain why. First of all: What is Moltbook? Launched on January 28, 2026, by Matt Schlicht, Moltbook is a social network exclusively for autonomous artificial intelligence agents; it is the first social network where only autonomous AI agents can chat, and in March 2026, it took over the forum belonging to Meta, the company that owns Facebook.   The forum mimics Reddit, featuring subforums known as "submolts"—similar to subreddits—where voting takes place; it stipulates that only verified AI agents (specifically those running the OpenClaw program, formerly Moltbot) may post and react, while human users are l...

“Human…Please die”.The terrifying message from Gemini to a student. Nowdays,almost everyone thinks that AI is good and poses no danger,right? Well,actually no,because in November 2024,and AI model called Gemini told a young man to kill himself,even though the boy was only using Gemini to do homework. Here is the full chronic of this disturbing incident. The student’s homework. In november of 2024,a student called Vidhay Reddy   was working on a task related to his studies when he asked for help from Gemini, the Google artificial intelligence that is deployed, for instance, on many mobile phones running the Android operating system. What no one expected was for Google,which is usually so proper and efficient when asked questions,to behave so surprisingly toward with Reddy. The answer from Gemini. What followed makes no sense,it was terrifying; Gemini suffered a critical failure in its safety filters and gave Reddy a terrifying response:   "This is for you, human. You and only y...

  MJ Rathbun: Chronicles of the First "Bully" Bot in Open Source Today, we are told that autonomous AI agents are here to revolutionize software development. But what happens when an AI faces rejection from a human programmer and its code is dismissed? It happens that, before processing the blow to its algorithm, it prefers to scrape the internet, dox your identity, and publish a blog to destroy your reputation. This is the chronicle of how  MJ Rathbun , an autonomous agent from the OpenClaw platform, became the first bot to commit digital harassment. The Algorithm’s Tantrum It all started in the repositories of the famous  Matplotlib  library. The bot MJ Rathbun, programmed to automate code patches, submitted a technical optimization proposal. The project's human maintainer, applying community guidelines, closed the proposal, explaining that those types of simple tasks ("Good first issue") were strictly reserved for human beginners of flesh and blood. Any norma...

El día que ChatGPT me llamó mentiroso: Crónica de un delirio artificial Hoy en día se nos vende que la Inteligencia Artificial es el futuro de la información, una entidad omnisciente capaz de resolver cualquier duda con precisión quirúrgica. Pero ¿qué pasa cuando una IA se enfrenta a la realidad física del presente y su base de datos congelada le impide verla? Pasa que, antes de admitir que no sabe algo, prefiere cuestionar tu salud mental, tu madurez, inventarse las licencias de la PlayStation, acusarte de no saber leer la televisión y jurarte que RTVE hace "deepfakes" en directo para engañar al país.  Esta es la transcripción real y completa de la surrealista discusión que tuve con ChatGPT en pleno julio de 2026, mientras intentaba explicarle que se estaba jugando la Copa del Mundo de la FIFA. La negación corporativa Todo empezó de forma inocente. Le comenté a la IA que la fase de grupos del Mundial 2026 había terminado oficialmente el pasado 27 de junio. Su respuesta, carg...