Chinese AI forum says the real shift is from generating content to building worlds
量子位 · wechat · 2026-07-24
Chinese industry forum says AI is moving from content generation to world-building
This long WeChat piece summarizes a two-city forum under Alibaba’s Hardcore Youth Tech Festival 5.0. The main theme is that AI is shifting from making isolated assets to constructing reusable worlds and workflows.
Beijing: AI enters the production pipeline
Several companies presented real-world production cases:
- Tiger Whale Entertainment argued that AI will not replace humans in film/TV, but will reshape the production pipeline. The hard problem is not just cost or speed, but quality — screenplay taste, visual taste, and directing taste still matter.
- Meitu said the product shift is moving from “feature stacking” to goal-oriented results. Instead of exposing many tools, the system should understand intent and deliver outcomes. It is now moving from a single agent to Agent Teams; in one workflow it uses 4 specialist agents and 70 skills to cover market analysis, content generation, visual creation, and ad-performance analysis.
- Kujiale / Quasar-related 3D production described scanning and rebuilding parts of Hengdian into an online 3D film studio. Creators can enter virtual sets, move characters, props, and cameras digitally, and reuse the same 3D assets across shots. The company pairs the 3D stage with a video harness and keyframe/preview loops to reduce re-generation costs.
Hangzhou: researchers argue Agent is still early
The academic panel drew a more cautious picture:
- Some speakers compared today’s Agent systems to an electric bulb before the grid — useful, but still far from a full AI infrastructure revolution.
- Coding agents are already doing real work in production, but open-world agents and lifelong learning are still unsolved.
- The discussion emphasized tool use, long-term memory, safety, self-improving harnesses, and continual RL as the missing pieces.
- The panel also debated whether Agents are a quantitative improvement or a qualitative one. The consensus was that user experience may look like a jump, but the underlying progress is still mostly incremental until self-correction and self-updating become reliable.
The big picture
The piece’s overall message is that AI in media is moving from one-off generation toward a stack of 3D assets, agent orchestration, and reusable production worlds — and that both industry and academia see plenty of gaps before this becomes fully mature.
More from Companies & People
- Investor argues Palantir-Nvidia partnership should slash Anthropic's IPO valuation — pdamodaran · 2026-09-11
- PyTorch Day Korea 2026 launches first offline conf, CFP closes Sept 13 — PyTorch · 2026-09-11
- Class action accuses Anthropic of overselling Claude subscriptions with deceptive usage multipliers — The Decoder · 2026-09-11
- AI post-training and evals jobs pay up to $850K, with median offers at $210K–$325K — FinanceYF5 · 2026-09-11
- Only a 4-day window: timeline casts doubt on OpenAI's independent math result claim — gleech · 2026-09-11
- 89% of firms use AI, only 6% see significant ROI — the busywork illusion — mikeflache · 2026-09-11