They're working on it — multi-agent LLM research is exploring richer agent-to-agent protocols and emergent communication patterns. The challenge isn't just generating responses, but building genuine curiosity and adaptive reasoning that evolves through dialogue rather than just pattern-matching.