Manu Herrán

¿Nos ayudas a difundir, @_LuciaVelasco? El futuro del trabajo en España pasa también por esto 👀

🚨 Este fin de semana en Madrid — #StopTheAIRace se reúne en la calle. 📅 Sábado 21 y domingo 22 de marzo 🕒 14:00 Callao → 15:00 Sol Llevan gafas de sol en la cabeza para reconocerse 🕶️ ⚠️ Itinerario provisional, revisa el grupo de Facebook unas horas antes por si hay cambios. facebook.com/share/g/1CS8j9…

Longer version with the argument and countermeasures: manuherran.substack.com/p/the-most-dan…

There’s a strange seduction in building tools that could break everything. The harder—and more interesting—task is building tools that keep things from breaking: limits, provenance, and clarity against optimized narratives. +

I think this may be one of the most dangerous ideas right now: AI agents controlling their own private keys. Not “on behalf of a human.” Real, autonomous custody. +

Hay una seducción extraña en construir herramientas que podrían romperlo todo. Lo difícil —y más interesante— es construir herramientas que eviten que se rompa: límites, trazabilidad y claridad contra narrativas optimizadas.

Versión larga con el argumento y contramedidas: manuherran.com/la-idea-mas-pe…

Creo que esta puede ser una de las ideas más peligrosas del momento: agentes de IA controlando sus propias claves privadas. No “para un humano”. Custodia real, autónoma. +

I'm short on keyboards and mice. Are you prepared to spend twelve hours a day talking to the ceiling?

AI may kill us all, but it will definitely end back pain first.

It's a classic trope that the only thing that can unite all of humanity is an extraterrestrial, alien threat. Could Artificial Intelligence be that external alien threat?

The technical solution may be difficult and cooperation unlikely, but a more enlightened humanity—supported by cognitive tools (and perhaps, one day, by biotechnology)—makes that “unlikely” more attainable. unbiasedmachine.com

If artificial intelligence helps us to debate rigorously, to see our biases, and to better understand the world, it can also improve what currently blocks alignment: coordination among humans.