In the field of AI, we are standing at a major turning point. Recently, at the 2026 Zhongguancun Forum Annual Meeting, Yang Zhilin, founder of Moonshot AI, made a shocking industry insight: starting from this year, the research paradigm of AI will undergo a qualitative change. AI will no longer be merely a helper to humans, but will gradually become the leader in scientific research.
Research "Digital Twin": Each Researcher Comes with a Large Number of Tokens
Yang Zhilin pointed out that future competition in scientific research will evolve into a deep integration of "computing power resources" and "innovative thinking." Under the new R&D system:
Task Synthesis: Each human researcher will be equipped with a massive number of AI Tokens (word elements). These "digital labor forces" can automatically synthesize new experimental tasks and simulation environments.
Autonomous Exploration: AI will take on high-level tasks such as defining reward functions and exploring new network architectures. This means that humans only need to set the ultimate goal, and AI can perform tens of thousands of iterations in virtual laboratories to find the optimal solution.
R&D Accelerator: This AI-driven closed-loop R&D model will shorten the speed of technological iteration from "months" to "days" or even "hours."
As an outstanding scholar graduated from Tsinghua University and previously worked at Google Brain, Yang Zhilin led Moonshot AI to achieve remarkable growth records.
Within just over two years of its establishment, the company's valuation has already exceeded $10 billion, becoming one of the most closely watched AI unicorns globally. Its growth rate even surpassed that of Pinduoduo and ByteDance during the same period. This dual recognition from capital and the market stems from continuous breakthroughs in long-text processing and fundamental large model technologies.
