标题： 从可解释到被解释的AI：关于证伪和量化解释的想法 显示英文标题

标题： From Explainable to Explained AI: Ideas for Falsifying and Quantifying Explanations

摘要： 解释深度学习模型对于医学影像分析系统的临床整合至关重要。 一个好的解释可以突出模型是否依赖于虚假特征，这会损害泛化并伤害一部分患者，或者相反，可能提出新的生物学见解。 尽管像GradCAM这样的技术可以识别有影响力的特点，但它们只是测量工具，本身并不构成解释。 我们提出了一种针对计算病理学中分类器的以人为本的机器-VLM交互系统，包括全幻灯片的多实例学习。 我们的概念验证包括（1）一个集成AI的幻灯片查看器，用于运行滑动窗口实验以测试解释的声明，以及（2）使用通用视觉-语言模型对解释的预测性进行量化。 结果表明，这使我们能够定性地测试解释的声明，并可以定量地区分竞争性的解释。 这为从可解释的人工智能到可解释的人工智能在数字病理学及其他领域的实际路径提供了可能。 代码和提示可在 https://github.com/nki-ai/x2x 获取。 显示英文摘要

摘要： Explaining deep learning models is essential for clinical integration of medical image analysis systems. A good explanation highlights if a model depends on spurious features that undermines generalization and harms a subset of patients or, conversely, may present novel biological insights. Although techniques like GradCAM can identify influential features, they are measurement tools that do not themselves form an explanation. We propose a human-machine-VLM interaction system tailored to explaining classifiers in computational pathology, including multi-instance learning for whole-slide images. Our proof of concept comprises (1) an AI-integrated slide viewer to run sliding-window experiments to test claims of an explanation, and (2) quantification of an explanation's predictiveness using general-purpose vision-language models. The results demonstrate that this allows us to qualitatively test claims of explanations and can quantifiably distinguish competing explanations. This offers a practical path from explainable AI to explained AI in digital pathology and beyond. Code and prompts are available at https://github.com/nki-ai/x2x.