Zhang Yiming on ByteDance AI Internally: Catching Up with OpenAI and Anthropic Will Take Until 2029 or 2030 at the Earliest
At an all-hands meeting for the AI team, Zhang Yiming said ByteDance is capable of competing with OpenAI and Anthropic, but truly catching up with frontier technology will take until 2029 or 2030 at the earliest, and its large models must benchmark against the international first tier. Foreign media reported that ByteDance's Seed team is pretraining a model with a parameter scale of up to 10 trillion, an order of magnitude comparable to Anthropic's most advanced model. Zhang Yiming previously emphasized at internal meetings that model development must adhere to long-termism and delayed gratification.
Why read this
AI score 58
Zhang Yiming gives ByteDance's timeline for catching up with frontier models and reveals the order of magnitude of the parameter scale of the Seed team's pretrained model.
Topics
Body · AI translation
Article body
On November 12, The Paper learned from multiple informed sources that ByteDance founder Zhang Yiming recently discussed the AI business again in an internal sharing session.
It is understood that at a recent all-hands meeting for the AI team, Zhang Yiming said that ByteDance is capable of competing with OpenAI and Anthropic, but that truly catching up to the frontier technology will take "until 2029 or 2030 at the earliest." ByteDance's large models should benchmark against the top international tier.
Foreign media reported that ByteDance's Seed team is pre-training a model with up to 10 trillion parameters, a scale comparable to Anthropic's most advanced models.
This is not the first time Zhang Yiming has internally focused on the development path of the AI business. Previously, The Paper learned from ByteDance insiders that Zhang Yiming said at an internal meeting that in building models, one should adhere to long-termism and delayed gratification, rather than using others' outputs to trade for a momentary leaderboard ranking.
(Source: The Paper) Middle of the article Embedded advertisement