Self-Healing with AI-Powered Retry Mechanisms

Is there ongoing research focused on implementing AI-based mechanisms within the Temporal orchestration engine that can dynamically adjust retry logic in real-time? Specifically, I’m interested in how these mechanisms could leverage context, past failures, and evolving conditions to optimize retries.