English(EN) Claude’s Protein Design Hit Rate Was 26.8%. One Target Returned 0 for 90.

Claude 的蛋白质设计命中率掩盖了靶点间的极端差异

作者 PulseAugur 编辑部 · [1 个来源] · 2026-09-01 06:30

对 Anthropic 的 Claude 蛋白质设计活动的分析显示，尽管总体命中率为 26.8%，但这一数字掩盖了不同靶点之间存在的显著差异。一个特定靶点在九十次尝试中均未成功设计，这与其他显示更高成功率的靶点形成了鲜明对比。分析表明，平均命中率并不能代表单个新靶点的成功概率，这表明 AI 预测存在高度的噪声和未经校准的置信度。 AI

影响强调了仔细解读 AI 绩效指标的必要性，尤其是在方差可能至关重要的科学应用中。

排序理由对特定 AI 模型在科学任务上表现的分析。[lever_c_demoted from research: ic=1 ai=1.0]

在 Towards AI 阅读 →

AI 生成摘要 · Google Gemini · 来自 1 个来源。我们如何撰写摘要 →

本文如何被排名

Signal score: 44 / 100; Composite score across the factors below. Higher = stronger signal that this story matters right now.
Newsworthiness bucket: Tool; 对特定 AI 模型在科学任务上表现的分析。[lever_c_demoted from research: ic=1 ai=1.0]
Source corroboration: Single-source cluster; Only one publisher covered this so far. Single-source stories can still rank when the publisher is high-authority, but they lack cross-source corroboration.
Topics: model release, product; Editorial topic classification. Feeds into how the story surfaces on /topic/<slug> hub pages and into the per-entity coverage mix.
AI-industry relevance: High; Clearly on-topic for AI-industry coverage.
Story freshness: Breaking (< 6h); Fresh story with cross-source coverage still developing. Ranking may shift as more sources report.

完整方法见我们的编辑标准。

报道来源 [1]

Towards AI TIER_1 English(EN) · Decoding AI · 2026-09-01 06:30

Claude 的蛋白质设计命中率为 26.8%。一个目标 90 次尝试仅命中 0 次。

<h4>Anthropic’s August 18, 2026 campaign produced 354 binders from 1,320 designs. But 26.8% is a portfolio average across sixteen targets, not the probability your target works: published per-target rates ran from 80% on TREM2 to 0-for-90 on maltose-binding protein, and our Augus…

本文如何被排名

报道来源 [1]

Claude 的蛋白质设计命中率为 26.8%。一个目标 90 次尝试仅命中 0 次。

相关实体

相关话题