Tim took part in a panel on Responsible and Trustworthy AI at the Innovate UK BridgeAI Annual Showcase moderated by Chanell Daniels (Digital Catapult) and alongside David Barnard-Wills (Trilateral Research), Guy Gadney (Charisma.ai) and Prof Keeley Crockett (Manchester Metropolitan University).
With an audience of academics and businesses who were applying AI in diverse industries from creative production and entertainment, to train repairs, fisheries governance, and lawn care, thinking about where and how public engagement can and should contribute to trustworthy and responsible AI is, at first glance, a challenging task. However, through our panel discussions a number of key themes emerged:
(1) Consequence scanning - Keeley Crockett pointed to the value of consequence scanning as a practical tool for businesses of every size to think through the potential impacts of their technologies, and from there, to both understand affected communities, and shape engagement with them.
(2) Engaging workers - In early remarks Guy Gadney reflected on the impact of the writers strikes in Hollywood on creative industry exploration of AI, and in discussion David Barnard-Wills noted the importance of employees feeling engaged in positive work, not least for retention of skilled staff. Embedding public engagement into commercial work needs to involve workers on a number of levels - from workers feeling empowered to express their own needs in relation to AI, to also having opportunities to be connected to voices from those affected by the products and services they are delivering.
(3) Meeting communities where they are - Keeley reflected on the importance of tailoring engagement to the needs of different groups, particularly marginalised communities - whose discourse might not be starting from AI and AI issues.
(4) It’s all about timing - Guy reflected that taking a responsible AI approach had led to some of their work slowing down at times. I reflected on the time commitments to public engagement, but also the opportunity to create more timely forms of public input - particularly if we invest in the ongoing infrastructures to support publics to be informed, connected and compensated for engaging, and that allow us to activate public input at the moments when it can have most influence.
(5) Put the product in front of affected communities - David described the importance of making sure product are ‘scale ready’ - avoiding leaving work on responsible AI and public engagement until its too late, and unanticipated issues of bias or other problematic impacts undermine a product. Keeley described how the Manchester People’s Panel for AI has brought businesses to pitch their products to publics who then have time to review and comment. And in our work building on the AI Fringe People’s Panel we’ve talked about the power of citizens review to pose questions and provide guidance on both policy and products.
Thanks to Chanell Daniels for moderating the panel, and inviting Connected by Data to take part. Channel was an observer for the People’s Panel on AI in London in November 2023.