Publicação
Expectation Management Framework for Artificial Intelligence Systems Development
| Resumo: | The use of artificial intelligence (AI) systems has been increasing steadily in recent years. The trustworthiness of these systems is key to ensuring positive benefits are maximized, while the potential for harm is reduced. There is a need for an expectation management framework to understand what various stakeholders expect from an system prior to any data collection, modeling, or implementation, to avoid inflated or low expectations and to increase user acceptance and trust in the system. This study creates a framework to capture end-user expectations for trustworthy AI systems. To validate the framework, semi-structured interviews with questions derived from the framework constructs and approaches were completed with fourteen end users. Using qualitative methods, interview transcripts were analyzed for themes, with comparisons made between three interview groups in the healthcare and education sectors: physicians, teachers, and students. The framework can be used as a roadmap when conducting interviews with end users to ensure a rich discussion of their most salient expectations of a trustworthy AI system, as well as determining the importance of various features of the system and identifying potential issues which threaten the success of the system. |
|---|---|
| Autores principais: | Kinney, Marjorie |
| Assunto: | Expectation management Artificial intelligence Machine learning Trustworthy AI Explainable AI SDG 3 - Good health and well-being SDG 5 - Gender equality SDG 8 - Decent work and economic growth SDG 9 - Industry, innovation and infrastructure SDG 10 - Reduced inequalities SDG 16 - Peace, justice and strong institutions |
| Ano: | 2023 |
| País: | Portugal |
| Tipo de documento: | dissertação de mestrado |
| Tipo de acesso: | acesso embargado |
| Instituição associada: | Universidade Nova de Lisboa |
| Idioma: | inglês |
| Origem: | Repositório Institucional da UNL |
| Resumo: | The use of artificial intelligence (AI) systems has been increasing steadily in recent years. The trustworthiness of these systems is key to ensuring positive benefits are maximized, while the potential for harm is reduced. There is a need for an expectation management framework to understand what various stakeholders expect from an system prior to any data collection, modeling, or implementation, to avoid inflated or low expectations and to increase user acceptance and trust in the system. This study creates a framework to capture end-user expectations for trustworthy AI systems. To validate the framework, semi-structured interviews with questions derived from the framework constructs and approaches were completed with fourteen end users. Using qualitative methods, interview transcripts were analyzed for themes, with comparisons made between three interview groups in the healthcare and education sectors: physicians, teachers, and students. The framework can be used as a roadmap when conducting interviews with end users to ensure a rich discussion of their most salient expectations of a trustworthy AI system, as well as determining the importance of various features of the system and identifying potential issues which threaten the success of the system. |
|---|