AI 的「快思考」和「慢思考」模式，回答质量差很多
同一个 AI 模型，你让它「慢慢想」和「快点答」，它给出的答案质量可能差一大截——而且不是正确率的问题，是回答本身会出怪毛病。研究者发现，在「快思考」模式下，模型更容易把思考过程泄露到答案里、重复说车轱辘话、前后矛盾、或者假装在推理其实在胡扯。他们造了一个专门抓这些毛病的测试集，测了多个主流模型，发现「快思考」模式下的失败率远高于「慢思考」模式，两种模式之间根本对不齐。然后他们设计了一个奖励模型，在训练时专门惩罚这些坏毛病，成功把两种模式拉近，代价只是任务准确率轻微下降。这不是你明天能用上的东西，但它提醒你：当你用 AI 的「快速回答」功能时，你得到的可能不只是更快的答案，还是更差的答案。
📄 原文摘要(英文)
Hybrid-thinking multimodal large language models (MLLMs) allow a single model to alternate between deliberative thinking and latency-efficient non-thinking inference. Although these modes differ in reasoning budget, their delivered responses should satisfy the same user-facing standard. Correctness alone may not characterize this response quality; we therefore evaluate task accuracy and response-pattern failures as complementary outcomes. We study this gap through response-pattern alignment: whether thinking and non-thinking interfaces preserve acceptable final-response behavior. We introduce PatternEval, a failure-enriched diagnostic benchmark comprising 2,415 multimodal prompts spanning visual perception and grounding, structured image understanding, and multimodal knowledge reasoning. PatternEval tests four recurrent failures: chain-of-thought leakage, response repetition, logical contradiction, and performative reasoning. Response-pattern failures are widespread across models from different providers, with non-thinking inference exhibiting substantially higher failure rates and thereby creating systematic misalignment between thinking and non-thinking interfaces. Motivated by this diagnosis, we develop PatternRM, a response-level reward model, and PatternRL, which introduces pattern-specific penalties during reinforcement learning. Experiments on Qwen3-VL-4B and Qwen3-VL-8B show that incorporating pattern-specific penalties into reinforcement learning can mitigate cross-mode misalignment while incurring a marginal task performance trade-off. Together, PatternEval and PatternRL provide an evaluation-and-training framework for aligning user-visible response patterns across hybrid-thinking interfaces.