header-langage
简体中文
繁體中文
English
Tiếng Việt
한국어
日本語
ภาษาไทย
Türkçe
Scan to Download the APP
Partnered With:

Meta loses another AI leader: Chain-of-thought core researcher departs to join $20 billion-valued Mercor.

Beating AI News Flash: Top AI researcher Jason Wei has announced his departure from Meta's Superintelligence Lab TBD Lab to join data startup Mercor as Chief Scientist. He had been at Meta for only about a year.


Jason Wei is a leading figure in the field of large model reasoning. He is the first author of the classic "Chain of Thought" paper, which advanced methods for enabling large models to solve complex problems through step-by-step reasoning. During his time at OpenAI, he participated in the development of the o1 reasoning model and Deep Research, and also led important evaluations such as SimpleQA and BrowseComp, which test AI's factual accuracy and deep search capabilities, respectively.


Mercor was originally an AI recruiting platform and now mainly organizes professionals such as doctors, lawyers, and engineers to create high-quality training data and evaluation questions for AI companies, helping models learn how to complete real work.


Mercor CEO Brendan Foody revealed that Jason Wei has been a client and exchange partner of the company for the past few years. Now, he will be responsible for building a research team and leading data and evaluation research. The team also includes Edward Hu, first author of the LoRA paper, who is in charge of model training.


Mercor just completed a $750 million Series D financing round on November 10, reaching a valuation of $20 billion. The company is expanding from supplying data to large model labs to helping more enterprises train and evaluate their own AI models.


Jason Wei believes that as more and more enterprises begin training models, data will determine the direction of AI's next stage of development.

举报 Correction/Report
RelatedRecommended
The market is reassessing the monetization exit of Meta's massive AI expenditure.
Meta is reportedly developing a standalone prediction market app, leveraging its massive user base. However, the biggest question remains whether to introduce cash transactions.
SemiAnalysis is bullish on Meta's catching-up speed, not its current model ranking
Before the launch of the cutting-edge model, the U.S. government needs to conduct a 30-day assessment, while Meta is still in negotiations.
Correction/Report
Submit
Add Library
Visible to myself only
Public
Save
Choose Library
Add Library
Cancel
Finish