AI and the Humanities
Georgia Wheaton interviews Baylee Brits
I had the pleasure of learning about ACU lecturer, Baylee Brits’, fascinating work in Artificial Intelligence, and how she used her literature degree to enhance AI for decision making.
Can you explain the project you’re currently working on?
I recently participated in a grant scheme run by the Defence Science Institute and Office of National Intelligence called 'Artificial Intelligence for Decision Making.' My project was on AI and deceit detection. I’ve since worked on broadening this project into an enquiry on AI and judgment.
How did you find yourself in this position?
I applied for an ‘AI for Decision Making’ grant and was successful. I thought that the project would prioritise computer scientists, and I was surprised and intrigued when I was told that the ONI / DSI were interested in different disciplinary perspectives and methods. I was keen to challenge myself by working on a (seemingly impossible) research question.
What kind of work do you do?
I conducted some experimental research on issues of textual interpretation in AI systems, using a General Imposters method to get a sense of the various possibilities and limitations of AI to anticipate or predict motivations and truthfulness in texts. I primarily studied persuasive/manipulative social media posts and whether they differed stylistically from persuasive factual content. I used this study to reflect on the prospects of AI ‘machine reading’ – can AI machines detect persuasive or manipulative language? What is the relationship between big data and deceit?
Do you enjoy working in this field?
Enormously. It is exhilarating to work in a rapidly evolving field with such social and technological significance. I enjoyed the challenge of having to learn skills and methods outside my usual discipline and speak to researchers with different priorities.
What is the intended outcome?
I suppose the aim is to provide literary / humanities-based insights into priority programs and 'wicked problems' for the ONI / DSI. And to extend research in the role of AI methods in textual interpretation.
What are the real-life applications of this research/work?
It depends on what you mean by ‘real-life.’ I think that this work has the capacity to modify or inform AI systems that rely on text interpretation – everything from chatbots to cyber security.
Can you speak on the intersection between humanities/literature and digital research in general?
That’s a very broad question. This can take on different forms depending on the interests and priorities of the researcher. I suppose -to give you quite a broad answer – this intersection essentially revolves around two things: the collision of quantitative and qualitative methodologies and the question of discernment. Techniques for analysing big data open up new questions and sources of information for research in literary studies – primarily in the form of quantitative data (think Franco Moretti). And literary studies – the traditional domain of interpretation and evaluation – can be vital to the growing reliance on ‘facts’ extracted from big data. Literary studies prioritise tropological, connotative and abductive methods for interpreting and appreciating texts, methods designed to analyse “rich-texts” which don’t necessarily mean what they say. I think humanities work is actually vital to the future of AI because it is focused on skills or attributes that currently stymie AI development.
The world of digital technology is broader than we think, as Baylee Brits has taught us, humanities, and literature specifically, can be paramount in the expansion of the artificial intelligence industry.