This paper contributes to a software engineering approach where AI technologies are carefully integrated inside critical interactive system to improve dependability. This property cannot be reached only using AI technologies as their probability of failure is usually high. But, in our approach, these technologies are introduced where the other technologies have reached their limit of dependability. To illustrate this approach, we use three different computer-based solutions for supporting post-stroke rehabilitation and demonstrates how mixing an AI component and a dependable component enables to improve both the dependability and the usability of the interactive system.

错误:搜索内容不能为空,请输入英文关键词
错误:关键词超出字数限制,请精简
高级检索

Adding a Non-dependable AI Component to Increase Dependability of an Interactive Application

  • Axel Carayon,
  • Camille Fayollas,
  • Celia Martinie,
  • David Navarre,
  • Philippe Palanque

摘要

This paper contributes to a software engineering approach where AI technologies are carefully integrated inside critical interactive system to improve dependability. This property cannot be reached only using AI technologies as their probability of failure is usually high. But, in our approach, these technologies are introduced where the other technologies have reached their limit of dependability. To illustrate this approach, we use three different computer-based solutions for supporting post-stroke rehabilitation and demonstrates how mixing an AI component and a dependable component enables to improve both the dependability and the usability of the interactive system.