Meta explains how AI operates in Facebook and Instagram, commits to openness and personalized experiences

0
56
Nick Clegg, president of Global Affairs at Facebook. Photo:Courtesy.
Share this

In a move to address concerns and give users more control over their content experience, Meta, the company behind Facebook and Instagram, is taking steps to enhance transparency and user customization. 

Via Metas official blog, President of Global Affairs at Facebook, Nick Clegg noted, “With billions of people relying on these platforms daily, Meta aims to ensure that the content displayed is relevant and interesting to individual users by utilizing AI systems.” 

The company claims it believes that openness and collaboration are key to responsible development in the rapidly advancing field of AI technology.

AI predictions and content recommendations

Meta’s AI systems employ a range of predictions to determine the value and relevance of content to users. By analyzing user behavior and feedback, such as post-sharing and survey responses, the systems make informed decisions on what content to display. 

Understanding the need for transparency, Meta is introducing “system cards” that provide accessible insights into how these AI systems rank content. 

A total of 22 system cards covering Facebook and Instagram are being released, offering information on various surfaces like Feeds, Stories, Reels, and more. 

The cards also shed light on AI systems that recommend content from accounts or groups not followed by users.

Furthermore, Meta is sharing detailed information about the signals, or inputs, that influence content ranking. 

These signals, which represent the majority of those used in Facebook Feed ranking, are now available in the Transparency Center. 

While Meta is committed to transparency, there are limits to the disclosure of signals that could potentially be exploited to bypass content safeguards.

Tools for personalizing user experiences

To empower users in shaping their content experiences, Meta has centralized control centers on both Facebook and Instagram. 

Through these control centers, users can customize their content preferences and influence what they see on each app. 

On Instagram, a new feature allows users to indicate their interest in recommended reels, enhancing the app’s ability to show them similar content. 

Additionally, the “Show more, Show less” feature on Facebook enables users to fine-tune their content consumption. 

Users can also opt for a chronological Feed on Facebook or select the Following tab on Instagram to bypass algorithmically-ranked content. 

Furthermore, favorites lists enable users to always see content from their preferred accounts.

Enhanced tools for researchers

Meta believes in an open approach to research and innovation, particularly in transformative AI technologies. 

In line with this belief, Meta has released over 1,000 AI models, libraries, and datasets for researchers over the past decade. 

This commitment continues with the upcoming launch of Meta’s Content Library and API, which will provide researchers with comprehensive access to publicly-available content from Facebook and Instagram. 

Qualified researchers pursuing scientific or public interest research topics can apply for access through partners experienced in secure data sharing for research purposes.

By introducing these tools early in the development process, Meta seeks constructive feedback from researchers to ensure the creation of effective and reliable research tools that meet their needs.

Meta’s endeavors to enhance transparency, user control, and research collaboration reflect their commitment to responsible AI development and their dedication to providing users with personalized experiences on Facebook and Instagram.

Share this

Leave a Reply