F. Javier Rascón | #Humanismo Digital en Salud

🚨 Stanford just discovered that AI might be making humanity dumber at diversity of thought. Not because the models are bad. Because they're all thinking the same thing. The paper is called "Artificial Hivemind." And once you understand what they found, you can't unsee it. Here's the breakdown: Stanford researchers built a dataset called INFINITY-CHAT. 26,000 open-ended questions. Creative writing, brainstorming, opinions, life advice. Prompts where there is no single right answer and diversity of response should be maximum. Instead they found the opposite. Two patterns kept showing up: Intra-model repetition - Ask the same model the same question across different runs. It keeps producing nearly identical answers. Inter-model homogeneity - Ask completely different models the same question. They converge on strikingly similar phrasing, metaphors, and reasoning chains. This isn't a benchmark problem. This is a culture problem. And here's why it happens: Most frontier models train on overlapping datasets. They're aligned using similar human feedback. They're optimized with similar reward models. So when you ask GPT-5, Claude, Gemini, and Llama to "write a poem about time" or "suggest creative startup ideas" or "give life advice"... they don't give you 4 different perspectives. They give you the same perspective 4 times, wearing different hats. The researchers call this the Artificial Hivemind effect. And the real horror isn't technical. It's sociological. Right now, billions of people use AI to brainstorm, write, think through hard decisions, generate ideas, draft opinions. If every model converges toward the same narrow band of ideas... We're not just homogenizing AI outputs. We're homogenizing human thought at civilization scale. The scary part the paper highlights isn't that AI surpasses human intelligence. It's that humans slowly outsource their thinking to systems all trained on the same worldview and nobody notices the compression happening. The internet was supposed to give us access to infinite perspectives. AI might be quietly reducing all of them to one.

‼️ATENCIÓN‼️: ❌ No es un meme ❌ No es IA ❌ No es El Mundo Today ❌ No es “Vaya semanita” ni “Pantomima Full” ✅ Son los Tribunales de Instancia de la fase 3. Sin duda: la mayor transformación de la justicia de la historia… para aniquilarla

El problema real es cuántas horas producen valor económico vs cuántas horas son teatro. Europa no está llena de vagos, está llena de trabajo de baja o nula productividad que sobrevive por inercia institucional. Podrías eliminar un % enorme de trabajos y nadie lo notaría. No porque la gente no valga, sino porque el sistema incentiva presencia, ritual y metodología… no resultados. Hay demasiado: * Roles cuya función es coordinar a otros roles que tampoco tienen output claro * Capas administrativas para cumplir procesos, no objetivos * Trabajo defensivo (reporting, compliance interno, reuniones) para cubrirse, no crear * Actividad confundida con impacto China (y EE. UU. en ciertos sectores) no gana por más horas. Gana por throughput, por eliminar lo que no escala, y por estructuras donde si no produces algo visible, desapareces. Europa juega a trabajar. Mucha profesionalidad performativa pero poca responsabilidad sobre resultados.

Hoy en Valladolid manifestándonos por la sanidad pública y contra la privatización salvaje que Mañueco lleva años perpetrando en Castilla y León. Frente al PP que destroza lo público, desde Podemos vamos a luchar junto a la gente hasta que la sanidad sea 100% pública y de calidad