Classification of Conversational Systems with Levels Based on “the Right to Speak''

DOI Web Site 4 References

Bibliographic Information

Other Title
  • 「発言権」に基づく対話システムのレベル分け
  • —A Proposal of a Milestone to Realize AI Accountability—
  • —AIアカウンタビリティの実現に向けたマイルストーンの提案—

Search this article

Description

<p>Arguably, one of the problems with artificial intelligence is its (un)explainability, and the so-called ``explainable artificial intelligence'' (XAI) is an allegedly promising approach to address it. There is also a trend against XAI: EU General Data Protection Regulation (GDPR) recognizes the right not to be subject to a decision based solely on automated systems for the reason that such systems are unable to handle the right to object. It seems that what GDPR asks for automated systems is a kind of functionality of conversational communication rather than mere explainability. Consequently, XAI would be insufficient. What should be addressed is the accountability rather than explainability of AI. However, an advanced conversational system that GDPR asks for would not be realized in the near future. In this paper, we argue that ``the right to speak'' could be a milestone to realize such a system and investigate how to implement the functions to handle the right. We also propose a classification of conversational systems with levels based on the right to speak, presumably equivalent to the SAE levels of automated vehicles. </p>

Journal

References(4)*help

See more

Keywords

Details 詳細情報について

Report a problem

Back to top