Cross-note synthesis

Range and diversity of participants

How the published AI CoLab notes, taken together, speak to the Range and diversity of participants evaluation theme. Public, general synthesis — safe to draw on without checking. It generalises the lessons; the sources below link back to the particular workshops (some members-gated).

56 insight notes so far. Not every workshop becomes one — recurring sessions and smaller gatherings usually don't. For some of the larger workshops a transcript is turned into an insight note so its lessons can be shared — with the room's consent, through time, and with people who weren't there.

The diversity that matters tracks the question being asked. The notes do not support "more diversity is always better" so much as match the mix to what you need made visible. Surfacing structural barriers took intersectional lived experience — gender, ethnicity, class, neurodivergence (Unlocking career potential). Stress-testing a working method took a cross-sector span of government tiers, academia and widely varying AI experience (AI for Insight Notes). Making adoption friction visible took a range from confident hands-on practitioners to people new to both AI and government (Growing with AI). Pressure-testing modernisation techniques took a cross-disciplinary mix — coders, designers and non-technical participants in parallel streams (AI for the Systems You Inherited). Rehearsing a high-stakes public failure took people carrying direct professional experience of a real welfare-automation collapse, set beside an international vantage on how other governments approach the same problem (When the Algorithm Goes Rogue). And working a shared ethical dilemma took the broadest span in the corpus — federal, state and local government, health services, cultural institutions, university research, private-sector practice, and human-research-ethics and First Nations perspectives — with groups deliberately mixed to maximise difference (Beyond the Principles). Scoping a shared data capability drew custodians, destination managers, AI startups and analytics consultancies across tourism, finance and agriculture portfolios (Tourism data); examining what AI change demands of a profession drew senior people-and-culture leaders from across the public service, private firms and academia (People and Culture Leadership). Demystifying the statistical foundations of AI, in turn, drew a deliberately broad span — federal agencies, consultants, startups and industry engineers, academia, postgraduate students and independents — in which a show of hands found social-science backgrounds in the majority with substantial engineering overlap, the very cross-disciplinary mix the session argued made deep learning possible in the first place (Beyond the Black Box). Scoping authoritative-data infrastructure for AI agents took a smaller but deliberately cross-sector room — private-sector legal-technology practitioners alongside public-sector regulatory, compliance and policy roles — which kept an abstract protocol anchored to concrete provenance obligations and civic, public-good uses (Authoritative by Source). And showing non-engineers that building is within reach took the span itself as the point — career engineers beside self-described non-coders, with neurodivergent perspectives explicitly in the room — so the contrast in confidence became the teaching material rather than a barrier to it (Vibe Coding for Curious Humans). Examining a high-risk system that judges people drew central agencies and line departments, independent consultants, former senior executives and early-career graduates, across disciplines from statistics and IT through governance, psychology and service design — the breadth letting the room test the system from the angles of those who build it, those who oversee it, and those who could be subjected to it (Getting Hands-On With High-Risk AI). And a strategic-foresight session made the experience spread itself the method, lining participants up from a few months to many years of AI exposure before mixing them across tables so no group was uniform (Beyond the Hype). Telling the story of crossing from pilot to production, in turn, paired a public-sector technology team with its private-sector orchestration-platform partner in front of a room of other agencies weighing the same step — a builder–vendor–peer mix that let the security, accuracy, identity and cost questions be pressed from each of those angles at once (From Pilot to Production). And opening up how AI is changing information work itself drew federal and state government, regulators, public servants turned founders, consultancies, industry and academia at very different stages of AI confidence — a spread whose value showed in the contrast between how the same people use AI inside government constraints and in their own ventures, set against international comparisons on adoption postures (AI and the changing nature of information work). And probing what agentic AI means for everyday work drew a room spanning public sector, community, industry and academia, whose range kept the conversation moving between front-line role impacts, organisational change, trust and governance (AI Productivity in the Agentic Era).

Other sessions in the corpus stretch the range in directions the APS-centred ones do not. The startup-and-SME session convened a deliberately non-government room — founders, SME leaders and not-for-profits alongside public servants — so commercial and social-sector stakes, not only policy ones, shaped what "responsible use" had to mean (AI Ready?). The human-AI teaming session made disciplinary spread the method itself, drawing anthropology, mathematics, physics, law and economics into one exercise on the premise that the cross-disciplinary mix was what produced the result (Human-AI teaming). And the environmental-stewardship session points to the vantage the corpus most often lacks — traditional land-management and cultural-knowledge holders set beside environmental scientists and AI technologists, where the explicit question was how to embed cultural knowledge in a technical workflow (AI for environmental stewardship).

Outside vantage points earn their place. A regulator from outside the sector sharpened the agriculture discussion (Growing with AI), and private-sector practice set beside public-sector delivery did the same in both 2026 notes (AI for the Systems You Inherited); an international facilitator gave a welfare-systems room a comparative window onto how several other governments are handling the same problem (When the Algorithm Goes Rogue). The outsider's questions are part of what makes the harder issues visible.

A gap that is starting to close: tracing what the mix changed, not just who was in the room. Mostly the effect of diversity is asserted rather than traced, but two notes show how to do better. Growing with AI shows newcomers grounding the discussion in real adoption friction; Beyond the Principles goes further and treats the mix as a method — one shared scenario drew out distinct health, recruitment, community-consultation and cultural-heritage readings because of who was present, and the divergence was the finding. People and Culture Leadership does the same with generational range, where the spread of personal stances on AI directly shaped the conversation about meeting different cohorts where they are. And naming who was missing is part of the same discipline: Tourism data flags that the room was mainly government and analytics-side, with private tourism operators absent — a gap that bounds what the session could conclude, and Practical AI for Policy People does the same from an introductory session, recording an APS policy audience across several agencies but noting broader cross-sector representation was thinner. The early corpus also shows the opposite of breadth, and naming it is part of the same discipline: some expert sessions were small or skewed toward a technical audience — a roughly seven-person discussion of multilingual model behaviour (How do LLMs behave in languages other than English?), and a developer- and architect-heavy room for an open orchestration framework (OpenSI-CoSMIC) — so their findings carry the weight of a narrow, specialist vantage rather than a broad one, a limitation to hold alongside the deliberately-mixed sessions above. Future notes should follow all three: name the insight a given vantage point produced, and the gap a missing one left.

Sources

Last compiled 2026-07-03 from 56 published note(s).

Notes contributing to this theme:

Shared with the AI CoLab Alliance — see join.aicolab.org.
This note was synthesised from the workshop recording with AI assistance and human review. It captures general lessons only, is not attributed to individual participants, and was reviewed for sensitive content and shared with the presenters before publication.