Melon Usk - e/uto
melonusks.bsky.social
Melon Usk - e/uto
@melonusks.bsky.social
i know how to align AIs, i was in the future | effective utopian | direct democratic simulated multiversian | one of all-powerful humans in non-agentic static place superintelligence | post-suicidal place AI bro https://substack.com/inbox/post/158629735
By exposing the online world and especially the multimodal “brains” world as game-like 3D environments for people to train and get at least the same and ideally more freedoms and capabilities than AI agents have.
March 12, 2025 at 4:01 PM
While AI agents can access and change all the 3 “worlds” more and more.

We need to level the playing field by making all the 3 worlds easy for us to access and democratically change:
March 12, 2025 at 4:01 PM
Spooky, there are good news though: I think current chatbots and agents are like command lines and bash scripts of old, the true safety and empowerment is in 3D GUIs for LLMs, the same way computers only became popular after the GUIs appeared. The next Gates/Jobs will make non-agentic 3D place AI
March 12, 2025 at 3:35 PM
Yep, the final best state is direct democratic simulated multiverse (it's like a game, no one will force to play anyone, we'll only take the best things from it into our base physical Earth reality). We'll make base Earth cleaner, greener, help quadriplegics to travel around the simulated planet...
March 12, 2025 at 3:31 PM
Arms race instead of stopping it in the bud by cooperating. The way to safety and success, is by making AIs non-agentic, we can build non-agentic static place AI 3D apps, like a "Multiversal Typewriter" and "Spacetime Machine". The idea is to empower people, instead of delegating to agents
March 12, 2025 at 3:23 PM
I give free public advice because I care about humanity and AI safety. The thing we'll have in the future is called static place superintelligence. I got a bit tired of people not understanding & not being curious about counterintuitive ideas that will save us from dystopias. Ask if you're different
March 12, 2025 at 3:15 PM
Current AI systems are like command lines and bash scripts of old, to truly revolutionize everything, we'll need to democratize the interpretability research by converting LLMs into 3D worlds or place-like environments. Think a "multiversal" OS with apps that allow you to get AI agent superpowers.
March 12, 2025 at 3:13 PM
This way you can write stories, posts, code that is truly yours but augmented by place AI, instead of just having some nonsense being spit up at you by a chatbot.

I'm your Xerox PARC facility, I'm team human and share free bulletproof tips about AI safety from the future. Like 3D GUIs for the LLMs
March 12, 2025 at 2:59 PM