Agentic-imodels: Evolving agentic interpretability tools via autoresearch

ArXi:2605.03808v1 Announce Type: cross Agentic data science (ADS) systems are rapidly improving their capability to autonomously analyze, fit, and interpret data, potentially moving towards a future where agents conduct the vast majority of data-science work. However, current ADS systems use statistical tools designed to be interpretable by humans, rather than interpretable by agents. To address this, we