From Audience to Investigators: Community Forums on AI

Public conversations about artificial intelligence often mix factual claims with disagreements about values, risks and priorities. From Audience to Investigators gives community members a practical way to examine those claims, identify trustworthy information, and discuss how AI should - or should not - be used.

The project is led by the Michigan Institute for Data & AI in Society (MIDAS) in partnership with Data Driven Detroit and participating Michigan communities.

More than listening to experts, participants will help define the questions, decide what evidence matters, test selected AI claims in real time, and identify safeguards and actions that institutions should consider.

Four community forums in Ann Arbor and Detroit will explore artificial intelligence, enabling participants to learn and be heard. The outcomes of the forum will become the Michigan AI Public Agenda, a public report documenting what Michigan communities want from AI, the information and safeguards they would accept, and where they agree and disagree.

Audience

• General public, U-M community

What it offers

The program combines community co-design, hands-on AI testing, and facilitated public deliberation.

• Hear contrasting perspectives from community members, researchers, government representatives, and industry professionals.
• Contribute your own questions and claims about AI’s capabilities, limitations, benefits and risks.
• Separate factual claims that can be tested from disagreements rooted in values and priorities.
• Help determine what information is needed and which sources should be considered trustworthy.
• Watch a “scientist on site” test selected claims using U-M’s MIDAS AI Sandbox.
• Compare AI outputs for accuracy, consistency, bias, framing effects and refusals.
• Discuss what the results mean and whether they change participants’ perspectives.
• Identify transparency requirements, safeguards, and practical next steps for institutions using AI.

Each approximately two-hour forum will move through five stages: Claims, Information, Testing, Deliberation, and Action. The goal is informed judgment, not forced consensus or increased AI adoption.

Community forums

February - May 2027 – Four public forums will examine distinct AI topics selected with community partners. Two forums will take place at U-M and two in Detroit.

Each forum will focus on a consequential public question. A possible example is how cities should use AI in housing and civic planning, including decisions about public resources, privacy, bias, transparency, and environmental costs.

Claims that can be examined during the event will be tested live. Claims that cannot be tested in the room will be documented along with the information, evidence, and institutional transparency needed to evaluate them.

Program development

September 2026 to January 2027 – The project team will work with Data Driven Detroit and other community representatives to select topics, develop the claim-testing system, and create facilitation and accessibility protocols.

The approach will be tested through several pilot activities, including:

• A U-M faculty teach-in.
• Participatory events within the Data and AI in Society lecture series.
• The public-facing AI and Mental Health conference on Oct. 5, 2026.

Lessons from these pilots will shape the final forum topics, testing methods, and participant experience.

Impact on the public good

The forums will help participants evaluate AI more critically, communicate across differences, and connect public dialogue to community-defined action. The project will also produce:

• The Michigan AI Public Agenda, documenting community questions, trusted sources, acceptable safeguards, and areas of agreement and disagreement.
• Interactive forum briefings that allow the public to explore claims, evidence, tests, and results.
• A Community AI Dialogue Toolkit with forum protocols, facilitation guides, co-design methods, evaluation instruments, and public-facing AI testing resources.

These materials will enable nonprofit, public, and community organizations to adapt the approach for their own conversations about AI.