AI Safety and Society

An expert-focused resource-sharing workshop from REAL ML, Ada Lovelace Institute and Open Society Foundations

The first global AI Safety Summit was a catalyst for the REAL ML community to come together for a one-day workshop to share resources on approaches and interpretations of AI safety, including by identifying opportunities for collaboration and partnership to ensure that AI safety isn’t prioritised to the exclusion of other considerations, like societal benefit and public interest development, and of course accountability.

Participants at this event discussed the dominance of AI safety narratives and lobbyists in emerging regulation, the growth of trust and safety as industry approaches to dealing with risk of harm resulting from AI and algorithmic technologies, and considered to what extent the AI safety narrative works for social justice approaches.

REAL ML and the Ada Lovelace Institute held this event to:

1. Sharpen participants’ understanding of implications of an AI safety approach and general AI focus of the UK AI Summit, so that they may take knowledge and critiques to summit events the following week, and to their organisations and institutions
2. Create a trusting space to analyse why and how the safety narrative has gained prominence, and for civil society to develop plans for action and advocacy
3. Create space for civil society to articulate what international AI governance can achieve, and articulate any limitations of this approach
4. Support participants to inform their respective fields and organisations around the globe, as the AI safety trend grows.