How Community Notes Reduce Viral Misinformation | Keith Coleman, Jay Baxter | TED

Community Notes: Building a Better Informed World

Introduction to Community Notes

• Community notes aim to create a better-informed world by providing accurate information across the internet.

• An example of a community note highlights misinformation about the USS Lincoln, clarifying that an image was AI-generated.

Trust in Community Notes

• The detailed nature of community notes fosters trust among users from various political backgrounds, surpassing generic misinformation warnings.

• Contributors write these notes, which are rated for helpfulness by diverse perspectives before being displayed publicly.

Scope and Impact of Community Notes

• All posts, including those from official accounts and advertisements, are eligible for community notes, enhancing transparency and accountability.

• Notably, even high-profile posts can be corrected by ordinary users, demonstrating the power of collective input in shaping public discourse.

Origins and Development of Community Notes

• The concept originated during the 2016 election when it became evident that misinformation was rampant on social media platforms.

• Previous fact-checking methods were slow (2 to 4 days) and lacked trust; thus, a new approach focusing on user contributions was necessary.

Mechanism Behind User Trust

• Users trust community notes due to their open and transparent process; anyone can verify how decisions are made through accessible algorithms.

• The algorithm prioritizes agreement from individuals with differing viewpoints rather than relying on external authorities for validation.

Visualization of Note Effectiveness

• A graph illustrates that only neutral notes—those found helpful by people with opposing views—are shown publicly; polarizing notes do not appear on the platform.

• This moderation system ensures that biased or unhelpful content is filtered out effectively through community engagement.

Handling Pressure from Influential Figures

• There is no override mechanism for removing community notes; if someone disagrees with a note, they must address it through public discourse rather than appealing directly to company executives like Elon Musk.

Post-Noting Dynamics

• After receiving a note, posts experience significant drops in engagement organically as users become aware of inaccuracies highlighted in the corrections provided by community notes.

• Research indicates that repost rates decrease significantly after noting occurs (by about 50%), reflecting users' willingness to adjust their beliefs based on new information presented through these corrections.

Challenges with Niche Topics

• While community notes typically appear within hours, there may be delays in niche topics where initial surprising agreements are hard to establish quickly due to low engagement levels at first glance.

Addressing Manipulation Risks

Defense Against Gaming Attempts

• The surprising agreement mechanism helps defend against naive manipulation attempts but more sophisticated attacks require additional safeguards such as verified phone numbers and monitoring raider behavior patterns.

• Even though incorrect notes occasionally surface, they tend to attract attention quickly leading them to be rated as unhelpful and removed promptly.

AI Collaboration in Note Creation

Enhancing Speed Through AI

• To improve speed further while maintaining quality control over content accuracy , an open API allows contributors to develop AI tools that assist human authors in creating faster yet reliable community notes .

• Human feedback plays an essential role in training AI models so they can learn from mistakes made previously , resulting ultimately into better collaborative outputs between humans & machines .

Future Directions: Bridging Divides

Connecting Perspectives

• A pilot program aims at identifying ideas liked across different viewpoints , promoting common ground instead of division among users .

• By highlighting shared sentiments around controversial topics , this initiative seeks not only foster understanding but also encourage constructive dialogue among diverse groups .

Visionary Open Source Knowledge Engine

Concept of the Knowledge Engine

• The knowledge engine transforms polarization into a shared understanding, promoting community connection.

• It is open source and utilizes open data, allowing various platforms to integrate and learn from diverse streams.

• This technology aims to bridge communities by facilitating AI learning from collective inputs.

Application Beyond Social Media

• Envisioning its use in Congress, focusing on areas of agreement like immigration and taxes could lead to positive outcomes.

• Pursuing common ground can enhance public satisfaction with governance and societal direction.

Pro-Social Media Future

Introduction of Guest Curators

• Audrey Tong introduces herself as a guest curator for TED 2026 alongside Divia Sedarth, emphasizing their role in showcasing impactful work.

• They aim to foster dialogue around significant issues such as AI and democracy through curated discussions.

Importance of Interview Format

• The interview format allows deeper exploration of complex topics that short talks may not fully address.

• Focuses on training AI to understand diverse community perspectives, enhancing democratic engagement.

Shifting Perspectives on Information

Community Notes Approach

• Current strategies focus on preventing misinformation but also consider how to promote agreed-upon solutions online.

Data as Soil Metaphor

• The concept likens data to soil that nurtures understanding among communities, fostering growth rather than extraction.

• Emphasizes the potential for AI agents to support community loyalty and regeneration of deep understanding.