2025 Industry-Wide Forum on AI Agents
UPDATED MAR 17, 2026
Forum date: November 2025
Countries: India and United States
Representative Sample: 503
Guiding Questions: How should AI agents be used? Where do users see the most value in AI agents?
See Results report here: Stanford Deliberative Democracy Lab report
We conducted an industry-wide Community Forum bringing together members of the public to provide their considered feedback on the future of AI agents. In partnership with Stanford’s Deliberative Democracy Lab, in addition to Meta, the collaboration included leading firms like Cohere, Oracle, PayPal, DoorDash, and Microsoft. This first-of-its-kind effort between major companies underscores how firms across the industry are asking similar questions as it relates to governance of this technology. The forum engaged a representative sample of 503 participants from the United States and India who were surveyed and deliberated on product and policy decisions where there are difficult tradeoffs with no simple answers –such as their preferences for AI agent guardrails and user controls, among others.
The results highlight the importance of understanding the US and India public’s perspective as it relates to technology that is seeing such rapid integration and impact. Specifically, we found several key findings on how people feel about AI agents, including:
- Consent, Autonomy, and Human Oversight. Participants supported the use of AI agents when automating repetitive tasks or focused on lower-risk tasks. They expressed caution on using AI agents for complex high-risk medical or financial tasks. However, people were more open to agent autonomy if provided safeguards around privacy and user controls such as requiring final approval.
- Transparency. Participants consistently prioritized being informed when interacting with an AI agent.
- Culturally adaptive agents. Participants in both countries favored culturally tailored AI agents over standardized ones, and they preferred this tailoring to happen through direct user input rather than automatic inference from location or language data.
- Building Trust. Lastly, discussions underscored the importance of transparency and education to foster greater public understanding and trust of AI agents and their capabilities.
With AI Agents expected to take on more transactions and tasks on behalf of users, there is a growing importance for governance mechanisms that help companies to better understand what it is people expect from this technology. We believe public consultation is an important step towards developing this understanding and hope the industry-wide nature of this forum can serve as a blueprint for continued discussion and collaboration on these important questions the industry is collectively facing.
For more details, view the full report from our partners at Stanford University’s Deliberative Democracy Lab here.