Future of general LLM work (interp/inference/alignment) vs agentic/physical AI (VLA, multimodal) for career [D] | TickerVault