AI 自己给自己出题，练出了 34 分的进步
让 AI 变强，通常靠人喂更多数据、设计更多任务。这篇换了个思路：让 AI 自己给自己出题、自己批改、自己迭代。研究者把任务按「底层能力」重新归类——比如「看图说话」和「听声辨物」表面不同，但可能都需要「跨模态对齐」这一种能力。然后 AI 针对每种能力生成简化但可验证的模拟世界来练习，练完再回到真实任务上检验。三轮角色循环：Creator 造题、Explorer 练题、Evolver 改题，28 个训练任务平均涨 34 分，18 个没见过的任务也涨 26 分，包括两个真实世界的机器人任务。它不是你明天能用上的产品，但它指向一个更省人力的 AI 进化方向：让 AI 自己当自己的教练。
📄 原文摘要(英文)
Multimodal intelligence can be viewed as human-guided co-evolution between worlds and agents through interaction. However, scaling this process faces two bottlenecks: constructing heterogeneous, verifiable environments is costly, and manual Harness iteration cannot keep pace with growing task diversity and complexity. A key insight is that tasks with different modalities and surface forms can share the same underlying capability requirements. We therefore introduce The Trinity, a recursive self-improvement framework coupling world generation with Harness evolution through three roles: Creator, Explorer, and Evolver. Drawing on verified execution trajectories, Evolver shifts the basis of task organization from surface categories to shared requirements, reorganizing the original task tree into a hierarchical capability tree. We introduce Capability-Driven Real2Sim2Real: for heterogeneous tasks at the same node, Creator abstracts their common requirements into simplified, verifiable simulated worlds for capability learning, while evaluation on original tasks measures transfer. Along this capability tree, Explorer progressively evolves reusable capability modules through selective inheritance, forming a Harness tree whose validated updates accumulate across iterations within a fixed topology. The resulting Harnesses specialize in different capabilities, motivating a Capability-Aware Harness Router that uses public task information to select compatible module configurations from the tree. Our evolved Harnesses improve mean performance by 34 points across 28 training benchmarks and 26 points across 18 unseen benchmarks, including two real-world embodied tasks.