Meta and Stanfords Deliberative Democracy Lab Release Results from Second Community Forum on AI Agents

In October 2024, Meta, in collaboration with the Stanford Deliberative Democracy Lab, conducted its third Meta Community Forum. This event built upon the October 2023 deliberations on Generative AI, focusing this time on two key questions:

1. How should AI agents provide proactive, personalized experiences for users?
2. How should AI agents and users interact?

Key Objectives of the Forum

• Global Participation: Expanded public input beyond the Global North to include ~1,000 participants from India, Turkey, Nigeria, Saudi Arabia, and South Africa.
• Advanced Topics: Shifted from foundational GenAI principles to specific value/risk tradeoffs (e.g., personalization, human-like AI).

Major Findings

Participants endorsed the following principles for AI agents:

• Memory and Personalization: Supported AI agents remembering prior conversations for personalized experiences, provided transparency and user controls exist.
• Cultural Tailoring: Preferred culturally/regionally-tailored AI agents over standardized ones.
• Human-Like Interaction: Favored AI agents capable of responding to emotional cues.
• Transparency and Control: Consistently emphasized the need for transparency and user control across all AI functionalities.

This Forum underscores Meta’s commitment to inclusive, public-driven AI development.