All the way from New York, Dr Meg Young breaks through the ‘participatory ceiling’ in inviting us to work on a new idea together before she’s even properly begun her talk in Canberra, Australia.
Calling on the expertise each of us are steeped in, Dr Young asks us to help road test the idea of a ‘subfloor’ as created by herself and co-authors Harini Suresh and Emily Tseng from Cornell Tech.
The idea of a ‘subfloor’ represents a meaningful participation system imbedded in use. In Dr Young’s world this meaningful participation revolves around Artificial Intelligence (AI). For those attending the talk either physically or online, this space held together was the system in which we were invited to deeply engage.
Meg wanted to put a problem space that she is trying to figure out. Inviting the room to “kick the tires” of this problem together and bring our variety of experiences with us, so that we could “get this thing right” together.
Foundation models are AI powered systems that train on the base of a large among of inputs such as images, text, speech, etc. and find patterns in the relationships of these. Using their ‘learning’, foundation models adapt what they’ve learnt to give results in many forms and varieties, such as answering questions, captioning images, or creating video. These AI systems are powerful, but the way that they are trained, built, and used currently perpetuates an incredibly large power imbalance.
Foundation Model diagram from ‘On the Opportunities and Risks of Foundation Models, Bommasani et al. 2021’
This is where participation comes in.
The ‘participatory turn’ approaches this power imbalance in foundation models, in designing and governing directly with those closest to impact there is hope that concerns can be mitigated and the power imbalance of the systems can be changed. There is a lot of work to do in this area, which is Dr Young’s role in New York at the non-profit Data and Society.
At Data and Society Dr Young works with focus groups to push through this ‘participatory ceiling’ in an interesting way. A process which is exciting, and new, and comes with many questions.
The beauty of interdisciplinary work, collaboration and community is that Dr Meg Young can bring this work-in-progress across the world and encounter a sweep of helpful questions, perspectives, and possible solutions. Bringing together people from different disciplines, cultivating community dialogues, and sparking interesting conversations feels incredibly right.
Cybernetics has foundational beginnings in bringing together the disciplines through the Macy Conferences 1946-1953. Indeed, Meg Young’s opening invite to work on this problem together holds historical echoes of the coming together of interdisciplinary community from the Macy conferences over 70 years ago.
This call to action did not go unheard, across the room throughout Dr Young’s presentation there was a large proportion of thought thinkers note takers, and question askers, followed of course by a roaring applause at the end of Meg’s talk. The crowd actively engaging in Dr Young’s work, ideas, and problem-posing continued the conversation in the kitchen, on stairwells, and in offices, long after the talk finished.
Dr Young has been delighted by the unique perspectives she has encountered here, enjoying talking to people who work with real world organisations, who have different applications from the front lines of their own research, and who offer takes on the ‘subfloor’ in respect to Indigenous communities. Highlighting the richness that interdisciplinary collaboration offers.
Following her talk, Dr Young is hosting an interdisciplinary session for PhD students across the ANU, attending a generative AI conference at ANU’s Kioloa campus, and will be attending AI events in Melbourne and Sydney.
Dr Meg Young’s talk co-hosted by ANU School of Cybernetics and ANU MINT Lab on 16 February 2024 along with her talk references can be found on our YouTube channel.
If you would like to continue the conversation with Dr Meg Young please reach out through her publictech space.
If you are interested in the work Dr Meg Young and her colleagues do at AIMlab, you can reach out to collaborate via email (aimlab@datasociety.net).