Tip
Liang Wenfeng Details DeepSeek's AGI Roadmap: Not Chasing Trends
In a recorded investor meeting, Liang Wenfeng outlined clear steps toward AGI: last year was Chain-of-Thought reasoning, this year is Agent, the next hard hurdle is continual learning, followed by the singularity (model self-iteration), and finally embodied intelligence. He views video generation and world models as side quests—getting the order wrong means more effort leads to more mistakes. Thus, DeepSeek does not chase trends but advances according to its own mainline rhythm.
Read the original (opens in a new tab)
News stream data aggregated by AI HOT