When we discuss with AIs we are fundamentally limited by information theoretical limits of our current language.
I was wondering if maybe LLMs are smarter than we think. Indeed in they have the ability to communicate complex concepts through the means of compressed latent representations.
Emergent language research looks at these questions and seems to be quite a rich field
When we discuss with AIs we are fundamentally limited by information theoretical limits of our current language. I was wondering if maybe LLMs are smarter than we think. Indeed in they have the ability to communicate complex concepts through the means of compressed latent representations. Emergent language research looks at these questions and seems to be quite a rich field