This talk aims to give a breadth-first overview of the Humanising Machine Intelligence (HMI) research initiative at the Australian National University (ANU), followed by a few deep-dives to inter-disciplinary research projects enabled or accelerated by this initiative. HMI was formed in 2018 as an intellectual collective to define and design moral codes for machine intelligence. In 2023, HMI comprises of 40+ faculty, postdoc, and PhD students across philosophy, computing, law and social sciences working on fundamental research, actionable solutions, and expert consultation for understanding, designing, and implementing computing systems in society. The talk will reflect on our intellectual and organizational trajectory and give an overview of the resulting interdisciplinary collaborations.
One project aims to understand the richness and nuances of daily moral dilemmas (over and beyond the trolley problem), using a corpus of over 100,000 real-life stories posted to the AITA subreddit. We design computational approaches to determine the topical landscape of dilemmas, and further quantify controversy, uncertainty in judgements made by reddit users. Another long-running research project focus on measuring and modelling online attention. We study how online items drive attention to each other and present new mathematical insights that connect microscopic actions by users and the implicit objective underlying content platforms. A third project is Smallset Timeline, a visual documentation tool (and associated software package) for data preprocessing decisions, represented as a “cartoon strip” of any changes made to a dataset. Tools like smallset address the challenge of provenance in machine learning pipelines and could be a plug-in to a growing set of documentation frameworks such as Datasheets and Model Cards.
About the speaker: Lexing Xie is Professor of Computer Science at the Australian National University, she directs the Humanising Machine Intelligence research initiative (https://hmi.anu.edu.au), and ANU Computational Media lab (http://cm.cecs.anu.edu.au). Her research interests are in machine learning, optimization, and AI in society. Of particular interest are stochastic point process models, neural networks for sequences and networks, applied problems such as modelling popularity in social media, vision and language, and understanding daily moral dilemmas. Her research is supported by the US Air Force Office of Scientific Research, Data61, Data to Decisions CRC and the Australian Research Council. Lexing received the 2018 Chris Wallace Award for Outstanding Research, her research has received seven paper awards and honourable mentions in ACM and IEEE conferences between 2002 and 2019. She was IEEE Circuits and Systems Society Distinguished Lecturer 2016-2017. She holds or has held a number of technical leadership and editorial roles, including ACM Journal of Responsible Computing, being inaugural editor-in-chief of AAAI ICWSM (Intl. Conf. on Web and Social Media), track chair at The Web Conference 2023 and others. She was research staff member at IBM T.J. Watson Research Center in New York before moving down under. She received B.S. from Tsinghua University, Beijing, China, and M.S. and Ph.D. degrees from Columbia University, all in Electrical Engineering.