Current limitations are number of tokens, but vector embeddings can potentially be used as kind of long term memory database for AI models. LLM can write their own code, search the internet.. it’s still early days and there are issues but the development is happening scarily fast
If you think @openai ChatGPT and @poe_platform bots are wild, check out Generative Agents and interactive simulacra. AKA interactive computational agents that simulate human behaviour. Welcome to @WestworldHBO
https://t.co/21g2vqYPJL
To understand Agents - they’ve only been around since the end of March - you assign the AI Agent a goal and it works away to solve it, spawning new agents to work on the problem. It continues to run until it finds a solution. Think Agent Smith (Hugo Weaving) in the Matrix.
I probably have mixed some links up in the previous post. Apologies! The images were correct though.
~
Nobles have voted and 4 winning Artworks have been bought! Congrats to the Artists :
1.@liamcampbell
2.@ArtFromAbove and @robwdawkins
3.@rs14213809
4.@liseykina