AI Recursive Self-Improvement
The cluster focuses on discussions about whether AI, particularly LLMs, can achieve recursive self-improvement by suggesting and implementing enhancements to itself, with debates on its feasibility, implications for AGI, and potential risks like rapid intelligence explosion.
Activity Over Time
Top Contributors
Keywords
Sample Comments
how do you know we're not at recursive self-improvement but the rate is just slower than human-mediated improvement?
Imagine the first llm to suggest an improvement to itself that no human has considered. Then imagine what happens next.
It's self-improving? So, we can ask AI how to improve AI, and the suggestions actually work?
It's begun. AI is improving itself.
The GP is talking about recursive self-improvement.What yes, it's clear by now it's way beyond the capacity of those AIs, and the odds are pretty good it's impossible to a large extent (but some limited version of it may be possible).
If the AI is actually human-level and is made by humans, then it will be capable of working on itself. Assuming the evolution of human intelligence was mainly limited by biological factors that don't apply to AI (like that brains can't be too big to go through the birth canal, that a body needs to optimize how much energy it gives to the brain vs other parts, etc), then if we hit human-level AI then we're going to be able to go past that. If humans are able to improve AI at any sp
A rudimentary form of self-improving intelligence :D
Those working on to improve the AI, until it needs no further improvement from humans because it can improve itself
Seems like a foreshock of AGI if the average human is no longer good enough to give feedback directly and the nets instead have to do recursive self improvement themselves.
Will a self-improving AI not do all the things you mention?