What Exactly Is RSI That VCs Are Chasing? Yao Shunyu, Shi Tianlin and Professors from Tsinghua and SJTU Weigh In
Summary
At a five-hour event in Shanghai's West Bund co-hosted by AGI House, Sequoia China and Research AI+, Google DeepMind's Yao Shunyu, Recursive Superintelligence co-founder Shi Tianlin (whose startup is valued at $4.65 billion), and researchers from Tsinghua and Shanghai Jiao Tong discussed the definition, evaluation and path to RSI (recursive self-improvement). The piece compiles eleven of their views on when RSI will arrive, where startup opportunities lie, and whether humans must stay in the verification loop.
Bubble analysis
This is mainly a compilation of technical opinions about how to define, evaluate and time RSI, with no investment figures, valuations or revenue data of its own. The only line touching the bubble question is the mention of a startup valued at $4.65 billion within six months of founding, which hints that capital is rushing into an unproven frontier — but the article never argues this point, so it bears only indirectly on whether AI is in an investment bubble.
Related signals
linked by shared tags
- CICC: Research Methods and Insights on AI Compute Hardware — Compute Chips, Storage, Equipment and Consumer Electronics, Sept 2026#research
- With most information hidden, the game Stratego had stumped AI—until now#research
- Photon held a funeral for mobile apps. Now it has $4.5M to help replace them with agents.#startup
- Ex-Qualcomm/Apple Engineers Team Up on WarpCore, Claiming It Will Upend Semiconductor Industry Rules#startup
- AI startup EliseAI valued at $4 billion in latest funding round#startup