A few months ago I asked why semantic representation rather than text wasn't used, since natural language seems quite a lossy representation for semantic concepts:
So the models will not only be using more and more Neuralese in their CoT, (like GPT-6), but different agents will also be able to communicate with each other in Neuralese. It's not looking good for monitorability.
https://news.ycombinator.com/item?id=47195212
I wouldn't have thought to use it for LLM-to-LLM communication, though