ABAB Pedia

Yuandong Tian

Yuandong Tian is an artificial intelligence researcher who was a research scientist director at Meta FAIR and is the first author of the paper on the open-source Go AI ELF OpenGo; he is a co-founder of Recursive Superintelligence.

Contents8 sections
Key facts

PhD from the Carnegie Mellon University Robotics Institute

Google Self-Driving Car Team

From 2013 to 2014 he worked on Google’s self-driving car team.⁠[1]

Paper on the DarkForest Go Program

In November 2015 he and Yan Zhu posted the paper “Better Computer Go Player with Neural Network and Long-term Prediction” on arXiv, describing a neural network–based Go program; it reported that the darkfmcts3 version reached a stable 5d level on the KGS Go Server and placed third in the January KGS Go Tournament.⁠[2]

Releasing ELF OpenGo

In February 2019 the paper “ELF OpenGo: An Analysis and Open Reimplementation of AlphaZero,” with him as first author, was submitted to arXiv and appeared in the proceedings of the 36th International Conference on Machine Learning (ICML). The paper describes ELF OpenGo as an open-source reimplementation of the AlphaZero algorithm and the first open-source Go AI to convincingly demonstrate superhuman performance, with a 20:0 record against top professionals worldwide, and says its code, models and self-play data were made publicly available.⁠[3][4]

Paper Awards

According to his personal website, his DirectPred work received an Outstanding Paper Honorable Mention at ICML 2021, and CompilerGym received a Distinguished Paper Award at CGO 2022.⁠[1]

Co-Founding Recursive Superintelligence

His personal website says he is a co-founder of Recursive Superintelligence Inc. and was previously a research scientist director at Meta FAIR, and that during his time at Meta GenAI he co-led reasoning work on Llama 4.⁠[1]

On May 13, 2026, GV announced that it was co-leading Recursive Superintelligence’s early $650 million funding at a $4.65 billion valuation and listed him among the seven co-founders alongside CEO Richard Socher. The company’s website says its aim is to automate knowledge discovery through recursive self-improving superintelligence.⁠[5][6]

Related Organizations, Websites, and Public Accounts

Carnegie Mellon University: Official website: https://www.cmu.edu/ (opens in a new window)⁠[1]

Recursive Superintelligence: Official website: https://www.recursive.com/ (opens in a new window)⁠[1][6]

Recursive Superintelligence: Official account: https://x.com/Recursive_SI (opens in a new window)⁠[6]

Sources

  1. Yuandong Tian (opens in a new window)
  2. Better Computer Go Player with Neural Network and Long-term Prediction (opens in a new window)
  3. ELF OpenGo: An Analysis and Open Reimplementation of AlphaZero (opens in a new window)
  4. ELF OpenGo: an analysis and open reimplementation of AlphaZero (opens in a new window)
  5. Recursive Superintelligence: self-improving AI (opens in a new window)
  6. Recursive — Self-improving superintelligence (opens in a new window)