The inner workings of the systems that decide the content users see on social media platforms Facebook and Instagram, both owned by the tech giant Meta, have been laid bare.
President of Global Affairs at Meta, Nick Clegg, released a comprehensive description of how these AI systems function, putting an end to user speculations regarding the nature of their feeds.
Meta’s Push for Transparency
According to Clegg, Meta is more upfront about the complex AI systems that consider users’ feedback while ranking content across social media platforms.
This disclosure forms part of a broader agenda aimed at promoting transparency, accountability, and openness, especially when advanced technologies like generative AI are involved.
Clegg said, “We believe that the best way to respond to those concerns is with openness…That starts with giving you more insight into, and control over, the content you see.”
This move comes as the European Union prepares to enforce the Digital Markets Act in 2024, which demands more transparency from online services about their recommendation algorithms.
The Inner Workings of AI
The AI systems of Facebook and Instagram determine the value of a piece of content for a particular user and decide its place in the user’s feed.
Factors like sharing a post indicate that a user finds that post interesting. Thus these actions play a crucial role in content recommendations.
It’s also emphasised that no single prediction can perfectly measure the value of a post to a user. Hence, the systems employ a combination of predictions, some based on user behaviour and some on user feedback gathered via surveys.
These prediction mechanisms, however, can also detect and curtail the spread of harmful content, in line with Meta’s Content Distribution Guidelines.
More Control for Users
Aiming to demystify its algorithms and give users greater control, Meta has released 22 system cards that reveal how the AI systems rank content.
This transparency aims to make the content ranking systems more accessible to users who aren’t tech-savvy.
The system cards cover various aspects, including Feed, Stories, and Reels, explaining how users’ interactions inform the machine learning models that deliver the content seen.
Meta is also introducing new tools and controls for personalising the user experience, including a new “Interested” feature being tested on Instagram’s Reels tab. This feature will allow users to see more of the content they like.
Clegg said, “You have the ability to shape your experiences on our apps so you see more of the content you want to see and less of the content you don’t.”
Support for Public Interest Research
As a testament to their commitment to openness and collaborative research, Meta will roll out a new suite of tools for researchers, including Meta’s Content Library and API.
These tools will provide researchers from qualified institutions access to data from public posts, pages, groups, and events on Facebook and Instagram.
Meta has partnered with the University of Michigan’s Inter-university Consortium for Political and Social Research for this initiative.
Meta’s decision to unveil the workings of its AI algorithms affirms its commitment to transparency and user control. While there’s an ongoing debate about the power and influence of AI, this disclosure represents a notable effort to invite users into the conversation.