YZi Labs(@yzilabs):“Data in robotics versus data in LLMs has a very different alpha decay problem.” — @tarunchitra, Managing Partner @robotventures Much of what LLMs learn can carry across contexts. Robotics data is far more conditional: what works depends heavily on the robot, the objects, and the environment. As robots move into more complex, real-world settings, previously collected data can become less relevant much faster. For builders and investors, that shifts the question from “How much data do you have” to “How quickly can you close the loop between deployment, new data, and adaptation”. From The Convergence: Bhutan by YZi Labs.

YZi Labs

@yzilabs

YZi Labs invests in ventures at every stage, prioritizing those with solid fundamentals in Web3, AI, and biotech.

参加 February 2018

385 フォロー中 756.5K ファン

YZi Labs@yzilabs

2026.09.23 14:08

“Data in robotics versus data in LLMs has a very different alpha decay problem.” — @tarunchitra, Managing Partner @robotventures Much of what LLMs learn can carry across contexts. Robotics data is far more conditional: what works depends heavily on the robot, the objects, and the environment. As robots move into more complex, real-world settings, previously collected data can become less relevant much faster. For builders and investors, that shifts the question from “How much data do you have” to “How quickly can you close the loop between deployment, new data, and adaptation”. From The Convergence: Bhutan by YZi Labs.