Explanation-Driven Interventions for Artificial Intelligence Model Customization: Empowering End-Users to Tailor Black-Box AI in Rhinocytology
Andrea Esposito, Miriana Calvano, Antonio Curci, Francesco Greco, Rosa Lanzilotti, Antonio Piccinno
TL;DR
The paper addresses the challenge of safely customizing black-box AI in high-stakes rhinocytology by introducing explanation-driven interventions within an End-User Development framework. It presents Rhino-Cyt, a CNN-based AI-assisted platform that allows rhinocytologists to edit AI explanations and adjust classifications, thereby guiding future model behavior without programming. The work provides a novel intervention-based UI, a principled design rationale, and a positioning within existing EUD taxonomies, highlighting how editable explanations can foster trust and a symbiotic human-AI relationship. The proposed approach has practical significance for clinical decision support and suggests broader applicability to other expert domains requiring human-in-the-loop refinement of AI reasoning.
Abstract
The integration of Artificial Intelligence (AI) in modern society is transforming how individuals perform tasks. In high-risk domains, ensuring human control over AI systems remains a key design challenge. This article presents a novel End-User Development (EUD) approach for black-box AI models, enabling users to edit explanations and influence future predictions through targeted interventions. By combining explainability, user control, and model adaptability, the proposed method advances Human-Centered AI (HCAI), promoting a symbiotic relationship between humans and adaptive, user-tailored AI systems.