“Governing (with) AI” Schedule

The event will be called to order with a friendly and fast-paced kickoff that includes words of welcome from the hosts, brief participant introductions, along with overviews of the agenda, participation guidelines and meeting logistics.

The program will begin with a series of interactive learning dialogues that explore selected facets at the intersection of AI and democratic governance. Participants will be invited to rotate between topics across the course of the session.

• The role of AI in deliberative democracy
• Can a chatbot facilitate deliberation?
• Using AI to synthesize a collective will
• The tools of governance (as opposed to government)
• Privacy and security issues in the use of AI in democratic processes
• Adversarial uses of AI in democracy
• Applying a power analysis to AI

The second half of the morning will focus on deeper discovery and problem articulation,
designed to inform subsequent strategy and design sessions.

• Mapping use cases for AI in citizens assemblies
• Inventorying the bugs of citizen assemblies that AI might solve
• The advantages and disadvantages of AI facilitated deliberation
• Open problems in using AI tools to connect citizens’ assemblies to the macro-public
• Democratic affordances: What does AI today make possible for democracy that previous technological and other revolutions did not?
• Employing AI to enhance citizen participation in international dispute resolution on collective rights
• Barriers and opportunities for the use of LLMs and chatbots in democratic deliberation outside the English-speaking world
• Towards a threat model of AI in democratic contexts

Participants are encouraged to dine with those who they have not yet met or engaged.

This is the end of the overlap between the “Governing Citizens’ Assemblies” conference and the “Governing (with) AI” conference. Invited guests from both conferences are welcome to stay for lunch. Invited guests from “Governing Citizens’ Assemblies” are free to depart Yale.

As a bridge to strategy and design sessions, participants will be invited to propose compelling provocations and debatable assertions that probe and test the interplay between AI, governance and democracy.

These working sessions will invite participants to further explore both AI in governance as well as governance of AI. Topics have been identified during pre-event engagement and planning.

Session facilitators will briefly introduce the objective of each session, and participants may then elect to join the session of their choice. Report-backs will be done at the end of the session slot.

Sessions currently anticipated to be in this time slot are:

• Enumerating and comparing design patterns for global deliberation
• How is AI currently regulated at the global level and what are the insufficiencies of that approach?
• What has worked before to positively shape technological development?
• The Power Question: Who controls AI, who owns the data, who is connected to AI, who benefits and who suffers?
• Will the “average voter” be the primary point of failure in AI governance?
• What does an environmentally sustainable, citizen-centric, AI-augmented democracy look like?
• Pathologies of tech companies’ governance

The closing session will invite participants to weigh in on what has been most useful during the course of Day 1, and refine their goals and priorities for the agenda of Day 2.

The day will start with a summary of Day 1 outcomes and a Day 2 Agenda Overview.

These working sessions will be similar in structure to those from Thursday afternoon.

Some sessions will continue and build on work started during Day 1. Other sessions will introduce new topics and objectives. Many topics will be sourced from participants during the course of Thursday, but other potential topics have been identified from pre-event input:

• What can make AI regulation democratically legitimate?
• What kind of AI for what kind of democracy?
• How would a global citizens’ assembly on AI governance fit into the global web of international laws and institutions?
• What are the risks and merits of open versus closed AI?
• How do we include tech employees’ voice in high level decisions about AI development?
• Data representativeness and biases: what do we know and still not know?
• Where are we on inclusion?
• Envisioning a curriculum for citizen education on the opportunities and risks of AI
• Collaborative policy prototyping: How can we iteratively prototype policies so we are more certain of their efficacy?
• Considering a constitutional approach for global AI regulation
• What reforms to the internal governance of tech companies should take place to ensure that AI remains safe for humanity?
• Actionable guidelines for balancing AI-driven innovation with democratic governance
• Weighing risks and opportunities of seeking to regulate AI models on the basis of their potential risk
• The future of work: artificial intelligence, corporate governance and workers’ rights
• Real time assessment of the social implications and governance of AI as the technology side of AI advances.
• Drafting concrete best governance proposals
• Graphing the pros and cons of EU versus US regulatory frameworks
• How can AI enhance small-group and large-group deliberation
• How might we evaluate AI safety in complex society-in-the-loop systems?
• What’s different about AI compared to other big issues, like climate, where we’ve tried to have mass democratic dialogue?
• What has worked before to positively shape technological development?

The group will pause before the final session to take stock of the progress made to this point in the week and to inventory action items, next steps and other bridges to post-event collaboration.

Participants will weigh in on what has been most useful during the course of Day 2, share appreciations and bring the meeting to a close.