Animesh Garg on X: "Physical AI is making a classic Moneyball error by mis-pricing data. Optimizing for "cumulative operational hours" or relying on low-variance production telemetry is the robotics equivalent of chasing batting averages, when we should be isolating on-base percentage. Unlike https://t.co/m7DOwExiHd" / X
Physical AI is making a classic Moneyball error by mis-pricing data. Optimizing for "cumulative operational hours" or relying on low-variance production telemetry is the robotics equivalent of chasing batting averages, when we should be isolating on-base percentage. Unlike https://t.co/m7DOwExiHd
Animesh Garg @animesh_garg Physical AI is making a classic Moneyball error by mis-pricing data. Optimizing for "cumulative operational hours" or relying on low-variance production telemetry is the robotics equivalent of chasing batting averages, when we should be isolating on-base percentage. Unlike text, every useful hour of robot data is paid for. If you aren't calculating marginal utility per dollar based on data novelty and intrinsic dimensionality, you're allocating capital based on vibes. In physical AI, useful data is strictly resource-constrained, shifting the goal from maximizing comp
related reading
- Moneyball for Physical AIpraxiscurrents.substack.com
- how we accidentally solved robotics by watching 1 million hours of YouTube – atharva's blogksagar.bearblog.dev
- What if RL Environments Aren't Mispriced?benanderson.work
- AI Is Slowing Downwheresyoured.at
- Cheap RL tasks will waste compute | Mechanize, Inc.mechanize.work
- Navigating the High Cost of AI Compute | Andreessen Horowitza16z.com
- Random thoughts while gazing at the misty AI Frontierblog.eladgil.com
- The Robot Is Ready. The Factory Isn't.linkedin.com
- Muskan Kalra (@muskan_kalra24) on Xx.com
- Bessemer Predicts: Robotics and physical AI - Bessemer Venture Partnersbvp.com
- Nikolaus West (@NikolausWest) on Xx.com
- AI Is Really Weirdwheresyoured.at