DeepMind Co-Scientist 走出模拟，在真实实验室闭环验证
Impressive new paper from Google DeepMind.
做 AI for Science 的同学必看，Co-Scientist 首次在真实实验室跑通闭环，从材料合成到生物预测都有硬结果，值得精读论文并复现其可靠性模块。
Impressive new paper from Google DeepMind.
谷歌DeepMind一篇令人印象深刻的新论文。
(bookmark it)
（收藏起来）
It takes Co-Scientist out of simulation and into real-world experiments.
它将Co-Scientist从模拟带入了现实世界的实验。
A summary of the results:
结果摘要：
In computer science, it found an inference-time scaling architecture that beat six frontier models on HealthBench Hard and Professional under blinded physician review.
在计算机科学领域，它发现了一种推理时扩展架构，在盲审医生评估下，该架构在HealthBench Hard和Professional上击败了六种前沿模型。
The system designed a safe precursor route for MXenes and drove a semi-automated chemical vapor deposition reactor, producing a lamellar 2D material with structural similarities to the Ti3C2Tx lattice.
该系统为MXenes设计了一条安全的前驱体路线，并驱动半自动化学气相沉积反应器，生产出与Ti3C2Tx晶格结构相似的层状二维材料。
It also tailored growth recipes to laboratory constraints in minutes, enabling single-attempt growth of monolayer MoS2, MoSe2, and WS2. In biology, it predicted E. coli swarming phenotypes across inducer gradients from sparse imaging data, matching unpublished real-world measurements.
它还在几分钟内根据实验室限制调整了生长配方，实现了单次尝试生长单层MoS2、MoSe2和WS2。在生物学领域，它从稀疏成像数据中预测了跨诱导物梯度的大肠杆菌群集表型，与未发表的真实世界测量结果相符。
30 domain experts wrote 450 reviews on end-to-end generated papers, and the reliability modules reduced hallucination and plagiarism.
30位领域专家对端到端生成的论文撰写了450条评论，可靠性模块减少了幻觉和抄袭现象。
Paper: https://arxiv.org/abs/2608.26701
论文：https://arxiv.org/abs/2608.26701
Chat with Paper: https://academy.dair.ai/papers/co-scientist-runs-closed-loop-experiments-in-real-labs-2608.26701
与论文对话：https://academy.dair.ai/papers/co-scientist-runs-closed-loop-experiments-in-real-labs-2608.26701
更进一步：量化金融体系
看懂新闻只是起点——沿量化金融路径，把它变成能交付的工程能力