JAKARTA - Meta has finally let a lot of eyes peek and learn how companies use Artificial Intelligence (AI) to determine what content users see when users use Facebook and Instagram.
"We need to be more honest about how this relationship works and give you more control over what you see," said Meta Global Affairs President Nick Clegg on the company's official blog, quoted Friday, June 30.
In this case, Meta released 22 system cards to provide information on how the company's AI systems scale content, some of the predictions each system makes to determine what content is likely to be most relevant to users, as well as controls to help customize the experience on Facebook and Instagram.
For example, the system card about the Instagram search function explains how applications retrieve relevant search results based on user queries, evaluate each result taking into account user interactions previously.
Not only that, the system card can also apply additional filters and integrity processes to narrow the list before presenting it to users.
While in the system card for Facebook content rating, there are various factors that affect the priority of platform information.
The factors include aspects such as third-party fact-checking marks on posts, account engagement rates, and user interactions with those accounts.
To provide further details beyond what's published on the system card, Meta also shares the type of input known as signal and predictive model informed by this signal, to help determine what content users think is the most relevant of friends on Facebook.
"We are also using signals to help identify harmful content, which we removed once we realized it, as well as to help reduce distribution of types of problematic or other low-quality content according to our Content Distribution Guidelines," Clegg said.
In addition, Meta is also developing the Why Am I Seeing This?" feature on Instagram and Facebook, where users can click on one part of the content and see how information from previous activities informs machine learning models that support content suggestions.
Finally, Meta will begin rolling out a new set of tools for researchers, namely Meta Content and API bibliography. The library includes data from public posts, pages, groups, and events on Facebook.
For Instagram, it includes posts and public data from creator and business accounts. Data from the Library can be searched for, explored, and filtered on graphics user interfaces or via programmed APIs.
The English, Chinese, Japanese, Arabic, and French versions are automatically generated by the AI. So there may still be inaccuracies in translating, please always see Indonesian as our main language. (system supported by DigitalSiber.id)
