As demands for increased transparency in artificial intelligence continue to grow, Meta has taken steps to provide users with tools and insights into how AI influences content recommendations on its platforms. In a recent release, the social media giant unveiled nearly twenty explainers focusing on different aspects of Facebook and Instagram, including features like Instagram Stories and the Facebook news feed. These explainers aim to shed light on Meta's content selection process.
The move comes as global legislation on AI disclosure looms, potentially requiring companies to provide more concrete information about their AI technologies. Meta's "system cards" delve into topics such as how the company determines recommended follow accounts, the functioning of search tools, and the mechanics behind notifications.
One example highlighted is the system card dedicated to Instagram's search function, which outlines how the app compiles relevant search results, assigns scores based on user interactions, applies additional filters and integrity processes, and ultimately presents a curated list to the user.
In a blog post on Thursday, Nick Clegg, Meta's President of Global Affairs, emphasized the importance of openness as the company unveiled its latest disclosures. Clegg tied the move to the ongoing global debate surrounding the potential dangers of artificial intelligence, which range from the dissemination of misinformation to the proliferation of AI-enabled fraud and scams. He acknowledged that while powerful technologies like generative AI offer exciting possibilities, they also raise legitimate concerns.
The blog post further outlined the intricate workings of Facebook's content ranking system. It delved into specific factors that influence the order in which information is presented on the platform. These factors include third-party fact-checking flags, the level of engagement from the account sharing the content, and the user's previous interactions with that account.
Meta's recent explainers coincide with the introduction of new tools that allow users to customize the company's algorithms. For instance, Instagram users can now request to see more of a specific type of content, expanding on the previous option to reduce content exposure. Clegg highlighted that both Facebook and Instagram users can further personalize their feeds through a menu accessible from individual posts.
Additionally, Meta plans to facilitate research on its platforms by providing a content library and an application programming interface (API) featuring a diverse range of content from Facebook and Instagram.
The announcement from Meta comes at a time when European lawmakers are advancing legislation that would impose new requirements for explanation and transparency regarding the use of artificial intelligence. Similarly, US lawmakers have expressed their intent to tackle similar legislation later this year.