Meta, the parent company of Instagram and Facebook, is taking steps to enhance transparency and user control over its AI recommendation algorithms. In a recent blog post, Nick Clegg, the company’s president of Global Affairs, expressed the company’s commitment to providing clearer insights into how these algorithms work across different products. Additionally, Meta is introducing a new feature called “Why am I seeing this?” for Instagram and Facebook Reels, as well as Instagram’s Explore page, to empower users with a better understanding of the content displayed to them.
Enhancing Transparency and Control
Clegg stated, “Today, we’re building on that commitment by being more transparent around several of the AI systems that incorporate your feedback to rank content across Facebook and Instagram. These systems make it more likely that the posts you see are relevant and interesting to you.” Meta aims to provide users with greater control over the content they encounter, addressing concerns regarding the opacity of AI-driven algorithms.
“Interested” Feature for Instagram
To improve user experience, Meta is testing a new system on Instagram that allows users to mark posts as “Interested” to see more similar content. Presently, users can already reduce the frequency of certain posts by selecting “Not Interested” from the three-dot menu. Additionally, the company is working on making Facebook’s “Show more, Show less” controls more prominent, although the exact changes to the controls’ location have not been disclosed.
Explaining AI Systems Through System Cards
Meta has released 22 “System Cards” that shed light on various AI systems, including Facebook Feed recommendations, Facebook Group timelines, suggested people and groups, Instagram Reels recommendations, notifications system, and Stories AI. These cards provide examples to elaborate on these topics, ensuring users have a comprehensive understanding of the underlying mechanisms. However, Meta acknowledges that these cards may require some time to read and comprehend fully.
Transparency Center and Research Invitation
In their efforts towards transparency, Meta plans to provide information about signals that impact user recommendations in the Transparency Center. These signals include actions such as liking or sharing a post. However, certain signals that could be exploited to bypass the platform’s safeguards will not be disclosed. Furthermore, Meta is inviting researchers to explore its algorithms through a Content Library and API, which includes public posts, pages, groups, and events from Facebook, as well as public posts and data from creators and business accounts on Instagram.
The Impact of Large AI Models
Meta is also considering the development of AI models with “tens of trillions” of parameters, surpassing the scale of current models like ChatGPT and GPT-4. These larger models have the potential to derive extensive insights from user data. While this capacity presents opportunities, it also poses challenges in terms of user control over content recommendations. Clegg previously emphasized the significance of user actions alongside algorithms in content display. However, users are not always aware of the signals captured by platforms to train their AI-powered feeds. To address this, Meta is exploring the concept of offering users greater algorithmic choices and control.
Meta’s Ongoing Efforts
In recent years, Meta has faced scrutiny regarding its content recommendation practices. Revelations by Facebook whistleblower Frances Haugen highlighted the adverse effects of the company’s algorithms on public content, including politics and news. Instagram, in particular, was identified as potentially harmful to teenagers. Additionally, Meta faced legal challenges from the U.S. Justice Department regarding its ad-serving algorithms, which were accused of depriving users of housing opportunities.
Meta is actively working to address these issues and regain trust. In the company’s Q1 2023 earnings call, Mark Zuckerberg revealed that AI-aided Reels suggestions on Instagram led to a 24% increase in user engagement. As part of its efforts to improve, Meta aims to fill up to 30% of users’ feeds with algorithmic recommendations.
To further demonstrate their commitment to transparency and user control, Meta is urged to draw inspiration from decentralized platforms like Bluesky and Mastodon. These platforms have sparked conversations about granting users more influence over the algorithms that shape their feeds. By adopting similar practices, Instagram and Facebook can offer users better algorithmic choices and increased control.
In conclusion, Meta is taking significant strides towards explaining and enhancing its AI recommendation algorithms. By prioritizing transparency and user control, Meta aims to address concerns regarding content suggestions and regain public trust. The release of System Cards, the “Why am I seeing this?” feature, and the invitation for researchers to study the algorithms all contribute to Meta’s ongoing efforts to provide a more transparent and user-centric social media experience.
Source: TechCrunch
