Quite a bit for last weeks belated weeknotes….
Unconference Insights
Last Thursday we had the first edition of Engaging people on data and AI: An unconference on giving people and communities a powerful say on public sector technologies aptly held in a residents association building in Pimlico.
The day was packed with interesting discussions on public sector tech, and public engagement, although it was notable how even with many public engagement practitioners in the room, the discourse often was pulled towards challenges of the technology, rather than practices for centring public voice.
A few specific take-aways and (in some cases, still half-formed) reflections I picked up from sessions I was in, reviewing session notes, and joining corridor conversations:
-
How can we create specific moments for inclusive engagement in audit of AI? Alongside ongoing transparency requirements, and public involvement in oversight boards, can we create regular points in time when the doors are thrown open and affected communities specifically invited into reviewing how systems are working. For example, should we have an annual time-boxed equivalent of the local govt accounts audit period (https://researchforaction.uk/how-to-challenge-your-councils-financial-accounts) when publics can request evidence of how algorithmic systems are working?
-
Can we empower young people in discussions of ed-tech and AI without investing more in building decentralised tech? Evidence from past engagement tells us young people object to being commodified by technologies, while recognising that tools can bring both benefits and harms depending on how they are used. But, when the only tech on offer is from large and distant corporations, the opportunity to shape an educational technology landscape that does not commodify is limited.
-
How do we celebrate stories of change? Teams in the civil service commissioning public engagement and delivering change as a result of input often don’t get the kinds of recognition and positive reinforcement feedback that could help drive future engagement activities. When public trust in government is low, explanations of ‘you said, we did’ may not be given credence.
AI Standard Hub & Scaling Participation
Rewinding to last Monday, I kicked off the week heading to the QEII Conference Centre for the ‘AI Standards Hub Global Summit’, particularly for the discussions of the role of civil society in AI standard setting. Standards play a significant, if often hidden, role in operationalising ethical and normative principles: from standards on data governance for AI training data, to process standards that can inform procurement of AI tools in the public sector. Yet, as speakers noted, standards processes are often long, convoluted, and span beyond the funding cycles of civil society. As artefacts with a significant, but rarely acute or determining, influence on the outcomes of digitisation, standards engagement presents a collective action problem for localised civil society groups. Few local or topic-specific civil society organisations have adequate incentive, or fundraising capability, to engage in-depth with global standards processes, let alone facilitate broader public engagement and inputs. Whilst technology offers no quick-fix, I did muse on BlueSky as to whether new tools could help us scale-out engagement with standards processes, addressing some of these structural challenges.
I’d also not clocked, until arriving at the venue, was the AI Standards Hub Summit this was taking place as a fringe of the much bigger Turing AI UK Conference. So I also took advantage of the opportunity to tour the stands and demo displays filling the venue. Striking that, although public engagement was on the AI UK programme at points, in particular, with presentation of the fantastic Children’s AI Summit, I didn’t spot a single display booth or stand communicating work on public attitudes to AI, or on how to carry out engagement with affected communities. Which links to a theme I’ll come back to: public engagement in technology governance is rarely argued against, but it also remains mostly invisible in the mainstream and technology development discourse - and that’s something I feel we need to think more seriously about.
Safe AI
On Monday afternoon I had the pleasure of joining a roundtable on the SAFE AI project, exploring potential for participatory engagement in shaping of humanitarian responses to AI. Notes from my inputs are written up here.
We & AI Celebration
On Wednesday I had the pleasure of joining an online celebration of We & AI’s fifth Birthday. It was fantastic to see the community that Tania Duerte and colleagues have built around AI literacy building and the Better Images of AI project, which, as it happens, we’re drawing on in Public Voices in AI People’s Advisory Panel workshops later this week.
The ecosystem of participatory digital governance
I was also working last week on trying to pull together early findings from our mapping work for the Open Government Partnership on participatory governance of digital technologies.
Whilst I’ve gathered a list of 70+ organisations across the world working, in some form, to support reforms that would see democratic public input shaping the design or regulation of technologies in public or private sector, I’ve found it much harder to source concrete examples and case studies that really show participatory governance of digital technology in practice. The best cases I can find seem to be at the municipal level, where there perhaps a tighter loop between participatory creation of guiding frameworks, and their implementation.
With data that is, right now, firmly resisting neat categorisation, I’ve been sketching out different ways of representing the different forms of participatory digital governance I’ve encountered. The first draft sketch is below
Working through this with designer, Virgil, has suggested the possibility of looking more at the relationship between different kinds of participatory practice: as in how, for example, participatory policy making can create the right environment for co-design of public services and pressures for public interest alignment of foundational models. More to do on this in the week ahead!
PAIRS Africa
Last, but not least, in updates, I’m looking forward to PAIRSx Africa on the 31st March: organised by Connected by Data Fellow Kiito Shilongo as an offshoot from the Participatory AI Research and Practice Symposium that took place in Paris in Feb. Do register for a set of fantastic presentations and discussions on ‘Pioneering new approaches to AI through Participation’ in Africa.