Meta Meets User Concern About AI Recommendations With “Openness, Transparency, And Accountability”

Meta is now trying to be "open, transparent, and accountable" for the way it pushes and recommends content to its millions of daily users. The company has tried to explain the logic and reasoning of the AI systems that power the backend.

Facebook and Instagram users have long wondered, and many-a-times, concerned about the way content surfaces on their walls and in their feeds. Meta appears to be taking a proactive step in addressing commonly asked questions about the algorithms that are responsible for content recommendation and flow.

22 "System Cards" Govern Content In The Feed, Stories, And Reels

Meta published a blog post this week, which is essentially an explainer of the ways Meta-owned social media platforms such as Facebook and Instagram recommend content to their users.

Meta's President of Global Affairs Nick Clegg is calling it an info dump on the AI systems behind its algorithms is part of the company's "wider ethos of openness, transparency, and accountability." He outlined what Facebook and Instagram users can do to better control what content they see on the platforms.

"With rapid advances taking place with powerful technologies like generative AI, it's understandable that people are both excited by the possibilities and concerned about the risks. We believe that the best way to respond to those concerns is with openness."

The majority of the mechanisms are governed by 22 "System Cards" that cover the Feed, Stories, Reels, and other ways that people discover and consume content on Meta's social media platforms. Each of these cards provides detailed information about how the AI systems behind these features rank and recommend content.

Instagram Explore, for example, is a feature that shows users photos and reels content from accounts they don't follow. The post explains the three-step process behind the automated AI recommendation engine.

1. Gather Inventory: the system gathers public Instagram content like photos and reels that abides by the company's quality and integrity rules.
2. Leverage Signals: the AI system then considers how users have engaged with similar content or interests, also known as "input signals."
3. Rank Content: Finally, the system then ranks the content from the previous step, pushing content that it predicts will be of greater interest to the user to a higher position within the Explore tab

Selecting or tapping options such as "Like", "Not Interested", or "Not Personalized" will influence, guide, and steer the algorithms. Meta's predictive AI models, which are the input signals used to direct the algorithms, and how frequently they're used to rank content, are explained in the Transparency Center.

New Content Recommendation Systems And User Actions Coming Soon

Meta is gradually expanding the "Why Am I Seeing This?" feature to Facebook Reels, Instagram Reels, and Instagram's Explore tab in "the coming weeks." This will allow users to click on an individual reel to find out how their previous activity may have influenced the system to show it to them.

Instagram is also testing a new Reels feature that will allow users to mark AI-recommended reels as "Interested". Needless to say, this will help the content recommendation engine pull similar content in the future.

The ability to mark content as "Not Interested" has been available since 2021, but several users have grumbled that this feature doesn't do much. It is possible that Facebook would improve this significantly in the future.