What Can Artificial Intelligence Learn from Medicine? Generative Analogies and Reliable Machine Learning Systems
Abstract
The paper explores the parallels between clinical translation in medicine and the development of machine learning systems, emphasizing the need for robust epistemic and methodological standards in ML.
Reality Card
The authors establish a generative analogy between clinical translation and ML system development, proposing a new form of ML reliabilism informed by the epistemic warrants of clinical translation.
The paper identifies specific epistemic and methodological warrants of clinical translation that can be applied to ML, enhancing the reliability of ML systems.
The paper does not provide empirical validation of the proposed analogy or its practical implementation in ML systems.
Paper to code
Verified implementation resources so builders can test the paper’s claims instead of stopping at the abstract.