Summary
Continuing to learn and get better over time is important for AI language agents, especially for complex professional tasks. The authors created a large test called Continual Learning Exam (CLE) to see how well these agents improve by gaining experience on multi-step workflows. The test includes 600 levels across various jobs and subjects, requiring agents to build on previous knowledge to solve harder tasks. They found that giving feedback helps agents improve, but even the best models struggle to fully master the tasks, showing there's still a lot of room to improve AI's long-term learning. This benchmark can also be used to train models to get better at learning over time.
What this means in practice
- •For llm developers: Improve training methods by using CLE to strengthen language models’ ability to learn from repeated experience in complex workflows.
- •For enterprise ai teams: Use CLE to evaluate internal AI assistants on sustained improvement in industry-relevant tasks before deploying in professional roles.
Authors
Haoyu Zhao, Sitian Tu, Yuxuan Huang, Huichi Zhou, Zhangyue Yin, Jiahao Huang, Jingyuan Xu, Tianmiao Shi, Jin Shi, Nianxi Zhang, Yangmiao Zhang, Zhuolang Chan, Huiyan Hu, Chenchao Sheng, Yaqing Luo, Yue Chen, Huaxiu Yao, Meng Fang, Jun Wang
Abstract
Continual learning is the ability of LLM agents to improve over time by acquiring, refining, and applying knowledge from experience while retaining useful prior knowledge. It is essential for agents learning over long time horizons, yet current evaluations largely assess individual task performance rather than improvement through accumulated experience. We introduce Continual Learning Exam (CLE), a large-scale benchmark developed by 50+ domain experts to measure experience-driven improvement in realistic professional workflows. CLE comprises 120 human-authored tasks, each organised as a five-level ladder, yielding 600 levels spanning 89 professional roles across 12 occupational domains. Agents must pass each level to unlock the next, with later levels building on experience accumulated during earlier assignments. An evaluation agent assesses each submission against human-authored rubrics. We compare a single-submission setting with one that provides diagnostic feedback after failure and allows one revision. Across 17 harness-model configurations using frontier LLMs, feedback and revision yield a median completion gain of 5.3 percentage points. Yet the best-performing system clears only 47.7% of levels and completes all five levels in just 23.3% of tasks, highlighting the difficulty of sustained learning from experience for existing frontier LLMs. As such, the dataset can also be used to post-train LLMs to strengthen their continual learning capabilities.