A carregar...
Publicação
From Explanation to Action
| Nome: | Descrição: | Tamanho: | Formato: | |
|---|---|---|---|---|
| 542.1 KB | Adobe PDF |
Orientador(es)
Resumo(s)
Agronomists bear the responsibility of verifying AI predictions to ensure safe agricultural decision-making. However, a critical gap persists in understanding how these domain experts leverage Explainable AI (XAI) to verify diagnoses within their workflows. We conducted a qualitative case study with nine professional agronomists to evaluate feature-attribution and example-based explanations. Our observations suggest that abstract heatmaps risk reinforcing automation bias by obscuring the biological ground truth. In contrast, example-based explanations appeared to support the experts' epistemic practice of situated seeing. Furthermore, participants consistently prioritized actionability over transparency. They viewed the diagnosis not as an endpoint, but as a prerequisite for intervention. Consequently, this work contributes design considerations to bridge the gap between static model explanations and active agronomic scrutiny.
Descrição
Publisher Copyright: © 2026 Copyright held by the owner/author(s).
Palavras-chave
Agronomy Digital Agriculture Domain Experts Explainable AI Human-centered AI Qualitative Study Human-Computer Interaction Computer Graphics and Computer-Aided Design Software
Contexto Educativo
Citação
Editora
ACM - Association for Computing Machinery