The Five Is: Key Principles for Interpretable and Safe Conversational AI

By Mattias Wahde et al

Published on Aug. 31, 2021

Read the original document by opening this link in a new tab.

Table of Contents

I. INTRODUCTION
II. BACKGROUND & RELATED WORK
III. DESIGN PRINCIPLES FOR TRANSPARENT CONVERSATIONAL AI
A. Interpretability
B. Inherent capability to explain
C. Independent data
D. Interactive learning
E. Inquisitiveness
IV. DISCUSSION

Summary

In this position paper, the authors present five key principles for the development of interpretable and safe conversational AI systems. The paper emphasizes the importance of transparency and accountability in the design and use of conversational agents. It discusses the drawbacks of black box approaches and advocates for alternative methods that ensure safety and transparency. The five key design principles proposed are interpretability, inherent capability to explain, independent data, interactive learning, and inquisitiveness. The authors provide detailed explanations and examples for each principle, highlighting their significance in creating transparent and accountable conversational AI systems.