OpenAI Dissolved Its Robotics Team Because Its Chatbot Model Can Now Drive the Robot. In an Nvidia Simulation It Beat the Purpose-Built Robot Models 75 to 51, and Detroit Teens Were Told the Same Week That the Trades Are 'AI-Proof'
About This Episode
Fortune reported on Friday, November 13 that OpenAI has dissolved its stand-alone robotics team and folded its people into the main research group under Mark Chen and the Applied organization under Greg Brockman; OpenAI confirmed it and says robotics remains core. The stated reason is that general-purpose frontier models such as GPT-6 Astra can now be applied directly to robots: in an Nvidia study an Astra-powered agent scored 74.8% on a simulated manipulation benchmark against 51.0% for the best specialized robot models. Anthropic made a similar claim for its own new model four days earlier, which matters for every worker who was told that hands-on work is the part AI cannot reach.
Our Take
The comfort for anyone who works with their hands was that AI is a mind with no hands and the hands would take much longer; this week OpenAI dissolved its robot team because it says the mind it already sells can drive the hands, so the episode weighs what a simulated 75 percent really proves against the 'AI-proof trades' advice Detroit teenagers heard the day before.
Continue Reading on Unscarcity
What's the Difference Between AI and Humanoid Robots?
Direct match: it explains why the robot body has lagged the AI mind (no internet-sized library of physical skill, so a body improves only as fast as someone can teach it), which is exactly the assumption OpenAI's reorganization bets against.
What Skills Will Matter Most When Robots Handle Labor?
Supplies the test for the listener's stake: career advice that names an activity ('learn to code', 'learn a trade') keeps expiring, and what holds value is judgment and being the person who answers for the result.