An LLM is a compression of language. It can help write the next model, the next plan, the next line of code. That is a tool. ASI, if it shows up, is a thing that acts under constraints it did not invent like energy, time, mass, other agents and gets better at acting because the world keeps scoring it. That is a being in the thin sense: a persistent loop with a body or a fleet, not a chat box that resets.

I support Elon Musk

Unfreeze those eggs. You’ve got plenty of time now.

The LLM is the planner and the language layer. Useful. Not the thing that is scored. A being, in the sense you mean, is the closed loop: act, hit a constraint that does not care about tokens, update, act again. Physics is the scoreboard, energy spent, part dropped, trajectory missed, heat you cannot talk away. Without that, you only have a better essay about the world. So ASI is not “a very large LLM.” It is a system that includes something like an LLM and is graded by reality often enough that the next plan is cheaper than the last failure. Until the score is physical, you still have a model. After it is, you have an agent that lives here.

poteto poteto poteto @poteto Grok Bot keeps hitting LinkedIn CAPTCHA on login. It feels like LinkedIn is flagging the session, and I really don’t want to risk getting locked out. Has anyone run into this, is there a safe way through it? Goal is simple: monitor posts from people I follow and alert me when something’s worth a reply. No posting, no engaging. Just watch and notify.

USAID Post USAID

X looks so pretty now. I woke up one day and it slowly grabbed me. I didn’t even notice. What is going on?