Google DeepMind 论文发现 AI 模型无法自主做出真正随机选择
New Google DeepMind paper finds that AI models cannot make truly random choices on their own, but larger models pick correctly when a computer program gives the
Google DeepMind 新论文发现，AI 模型无法自主做出真正随机选择，但较大模型在由程序先给出随机数时能正确选取。研究让 Gemini、Qwen3 和 OLMo 各从简单分布中采样 1024 次，所有匹配目标概率的统计检验均未通过，调整 temperature 或 top-p 也无法解决；在真实随机检查可让攻击者获胜 75% 的安全游戏中，Gemini 的选择让攻击者获胜率达到 96.6% 至 99.9%。作者建议智能体需要随机性时在代码中生成并传入，而不是让模型自己随机。
推荐理由
AI 评分 67
论文用统计检验和攻防游戏量化了模型无法自主产生随机性，并给出把随机数交给代码生成的工程做法。
主题
正文
New Google DeepMind paper finds that AI models cannot make truly random choices on their own, but larger models pick correctly when a computer program gives them a random number first.
Agents often need to pick actions at random with set odds, like checking servers in an unpredictable order. LLMs say they will pick uniformly, then keep landing on favorites like 7, 42, or answer C.
They asked Gemini, Qwen3, and OLMo models to sample from simple distributions 1024 times each. They also tried other temperature and sampling settings, and gave models a random number to convert.
Every reported statistical test for matching the target odds failed, and changing temperature or top-p did not fix it. In a security game where truly random checks let attackers win 75% of the time, Gemini's picks let them win 96.6% to 99.9%.
If your agent needs randomness, generate it in code and pass it in, instead of asking the model to be random.