One Brain, Any Body: Google DeepMind's Keerthana on Gemini Robotics 2, Cross-Embodiment & Humanoids

Google DeepMind's Gemini Robotics 2 adds whole-body humanoid control and multi-fingered dexterity, but generalist robotics remains at a "GPT-2" stage.

• Still GPT-2 era: Keerthana Gopalakrishnan rates robotics intelligence at a 'GPT-2 moment,' citing weak cross-embodiment generalization as the key gap versus LLMs.
• Gemini Robotics 2 launched: Three models released — ER2 (embodied reasoning, Gemini Flash-based, public API, ~128K context ≈3 minutes of memory), Gemini Robotics 2 (VLA, whole-body control from fingertips to feet), and an on-device model trainable on ~200 examples.

Themes, sentiment and reasons are AI-written paraphrases, checked so they never quote an episode. They may contain errors. Not investment advice. Episodes remain the property of their creators.