Towards trustworthy Human-AI collaboration in predictive visual analytics

PhD Qualifying Examination


Title: "Towards trustworthy Human-AI collaboration in predictive visual 
analytics"

by

Mr. Kam Kwai WONG


Abstract:

Human-AI collaboration capitalizes on human expertise and AI’s computation 
power to overcome weaknesses from both sides. Predictive visual analytics 
synergizes the collaboration with the human perceptual ability to promote 
better decision-making in high-stakes applications where predictive tasks are 
predominant. Nevertheless, when such systems fail to maintain appropriate 
trust, users become over-reliant to gain ill-informed insight or distrustful to 
limit the system’s adoption. How human-AI trust is built, enhanced, and eroded 
in predictive tasks remains an open question. This survey aims to identify 
influential trust factors and trust measurement in predictive visual analytics. 
A question-based framework is proposed to recognize the multifaceted 
contribution of recent advances and answer the “whats” and “hows” trust is 
investigated. Furthermore, the benefits and drawbacks of different trust 
enhancement and calibration approaches are discussed to provide insightful 
design implications for establishing an appropriate level of trust. Finally, 
this survey brings forward the challenges and future research opportunities 
towards trustworthy Human-AI collaboration.


Date:			Thursday, 2 December 2021

Time:                  	10:00am - 12:00noon

Venue:			Room 3494
 			(lifts 25/26)

Committee Members:	Prof. Huamin Qu (Supervisor)
 			Prof. Raymond Wong (Chairperson)
 			Dr. Hao Chen
 			Dr. Tristan Braud


**** ALL are Welcome ****