Choosing between LLM and agentic/physical AI research

The author is deciding between two research paths: LLM-centered work (alignment, interpretability, inference/optimization) which currently has more jobs and transferable skills, and agentic/physical AI (agents, multimodal, VLAs, robotics) which has fewer roles but growing investment and higher entry barriers. They ask about trajectories over 4–6 years and how transferable skills between the fields are.

• Hugging Face launches ToyBlocks robot-learning sandbox2026-11-06
• Community questions open-source models dropping security capabilities2026-11-06
• Hugging Face CEO says his P(Doom) is 'close to zero'2026-11-06
• Open ASR Leaderboard adds Egyptian and Nigerian English accents2026-11-06
• Hugging Face Releases MoE Training Handbook Built on Megatron-LM2026-11-05