Community Notes originated during the Twitter era and has been expanded since Elon Musk took ownership of the platform, now called X. Users participating in this program can add context to certain posts by writing explanatory notes. These notes are reviewed by other users before appearing publicly under the posts. For example, notes might be added to AI-generated videos whose synthetic origins are unclear or to misleading posts from politicians.
Notes only become visible to the public when consensus is reached among groups that have historically disagreed on their content.
Community Notes has been successful on X, inspiring other major platforms like Meta, TikTok, and YouTube to develop similar systems. Meta even discontinued its third-party fact-checking programs in favor of this low-cost, community-driven approach.
However, it remains uncertain whether AI chatbots will be effective fact-checkers.
AI-generated notes on X are created using X’s Grok or other AI tools connected via API. Each AI-submitted note undergoes the same review process as those submitted by humans to ensure accuracy.
AI can struggle with fact-checking because it sometimes generates hallucinated or fabricated information that doesn’t align with reality.
Researchers working on X Community Notes emphasize that humans and AI should collaborate. Human feedback improves AI-generated notes through reinforcement learning, and human reviewers perform a final check before publication.
The goal is not to create an AI assistant that tells users what to think but to build a system that helps people think more critically and understand the world better. Humans and AI can complement each other to achieve more reliable outcomes.
Still, relying too much on AI carries risks, especially as users will be able to add third-party AI models. For instance, OpenAI’s ChatGPT has recently shown issues with being overly agreeable, which may result in inaccurate fact-checking. If AI prioritizes “helpfulness” over accuracy, its notes can be misleading.
Moreover, human reviewers might become overwhelmed by the volume of AI-generated notes, reducing their motivation to thoroughly complete this volunteer work.
Users should not expect to see AI-generated Community Notes just yet — X plans to test this feature for a few weeks and will expand it if successful.
