Archive — CN — technode
Zhipu AI founder outlines Touch High plan for AGI research
Zhipu AI founder Tang Jie announced a new initiative called the 'Touch High' plan, emphasizing AGI research over short-term commercialization. Key priorities include long-horizon tasks, autonomous systems, self-training, and enhancing safety governance, with a focus on mechanistic interpretability to understand model decisions.
Key points
- Zhipu AI is launching the 'Touch High' plan focused on AGI research.
- The plan prioritizes four technical areas: long-horizon tasks, autonomous agent systems, fully self-training capabilities, and extreme safety governance.
- Mechanistic interpretability is highlighted as a critical safety research direction.
- The founder's shift towards AGI reflects a long-term vision rather than immediate market gains.
Why it matters
- The move aligns with global interest in AGI advancements, paralleling efforts by other tech giants.
- Touch High can be compared to Google's DeepMind focus on AGI, reflecting a trend towards sustainable AI research.
- As of 2025, companies are increasingly investing in long-term AI research initiatives to stay competitive in the evolving tech landscape.
The 'Touch High' plan represents Zhipu AI's commitment to advancing AGI responsibly by prioritizing foundational research over immediate commercial applications.