Impact of Capability on Trustworthiness of AI in a Chest CT Reading Tests
-
- TAKAGI Hisashi
- Gifu University
-
- YANG Li
- Nagoya University
-
- KOMARI Masatugu
- Osaka Electro-Communication University
-
- TERADA Kazunori
- Gifu University
Bibliographic Information
- Other Title
-
- 読影性能が胸部 CT 読影検査における AI の信頼性に与える影響
Description
<p>Trust in AI is affected by biases, and the impact of aversion bias in particular on trust is not well understood. In this study, we focused on algorithm aversion bias and examined the impact of disease risk on AI reliability in reading tests. Participants were presented with different levels of disease risk (10%, 40%, 60%, and 90%) and asked whether they would undergo additional testing if the AI and physician were readers of chest CT images. Results of the experiment showed that the magnitude of risk affected reliability, but not the reliability of the AI and the human physicians. This suggests that in situations where the magnitude of risk is explicit and no intentionality toward the patient is assumed, the reliability of reading is the same for human and AI physicians.</p>
Journal
-
- Proceedings of the Annual Conference of JSAI
-
Proceedings of the Annual Conference of JSAI JSAI2023 (0), 3I5OS4c02-3I5OS4c02, 2023
The Japanese Society for Artificial Intelligence
- Tweet
Keywords
Details 詳細情報について
-
- CRID
- 1390578283198030720
-
- ISSN
- 27587347
-
- Text Lang
- ja
-
- Data Source
-
- JaLC
-
- Abstract License Flag
- Disallowed