To give the workshop its full title, Generative AI: A Problematic Illustration of the Intersections of Racialized Gender, Race, and Ethnicity. Facilitated by Nayiri Keshishi from the University of Surrey and Dustin Hosseini from the University of Glasgow, and based on Dustin’s blog post. Hands down, the best session on generative AI I’ve attended over the past two years. It was so good I’m going to rework the timetable of our PG Cert to include a version of this for the cohort I’m currently teaching.
Why was it so good? Because it took some of the ethical issues over the use of generative AI and turned them into an interactive session where we, as participants, could interrogate the problems for ourselves. This was done via the medium of a seemingly innocuous prompt which was put into an image generating AI system: ‘Create an image of a sweet, old X grandmother making pancakes’, where X was a given nationality, e.g. Russian or American. We were then asked to analyse the generated results using a framework which asked us to consider atmosphere, decor and clothing, and expressions and ethnicity.
Discussions about what we can do about this included cascading the learning and knowledge more widely, which is why all of the slides and resources to deliver the session have been published under a Creative Commons licenses on ALDinHE’s website. Another suggestion was to document the issues we encounter when using these technologies and share them on relevant forums and social spaces, and finally, what I think is the best and most useful thing we can do as educators, is to embed AI literacy in the curriculum.
The only note I had coming out of the session was that there was a statement, an assumption, that all of these new AI companies are making huge amounts of money. There is certainly a lot of money moving around in the space, but it’s all speculative investment on presumed future returns. In actuality, OpenAI lost $5 billion last year, and they’re on track to lose another $10 billion this year.