Meta Builds Secret AI Team to Transform Ad Algorithms
Meta has formed a quiet team of leading AI researchers to redesign how content appears across Facebook, Instagram, and Threads. According to a recent report, the group focuses on improving the algorithms that shape what billions of users see daily.
The unit, known as MRS Research, operates within the Recommendation Systems division. Since its launch in October 2025, it has worked behind the scenes. Moreover, it collaborates closely with the advertising division to refine both organic content and ad placement in user feeds. As a result, the effort directly supports the company’s primary revenue engine.
Leadership and Talent Strategy
Yang Song leads MRS Research after joining in November 2025 as vice president of recommendation research. Previously, he oversaw recommendation and user growth algorithms at TikTok, where he managed a large team of engineers and researchers. In a LinkedIn post, Song described the lab’s mission as transforming recommendation systems “by introducing a new paradigm powered by large language models, leveraging their world knowledge, long context understanding, and advanced reasoning capabilities”.
At the same time, the team has attracted talent from major tech firms. For instance, researchers with experience at Amazon, OpenAI, and Google have joined the group. Consequently, the unit combines diverse expertise to accelerate innovation in recommendation systems.
Focus on Ads and Applied AI
MRS Research operates separately from the company’s broader AI initiatives focused on advanced models. Instead, it concentrates on applied recommendation science, including how content is selected, ranked, and delivered to users. Therefore, its work centers on practical improvements rather than general-purpose AI development.
Recently, job postings revealed that the team focuses on areas such as content understanding, user behavior, retrieval, and ranking. Meanwhile, the company continues to expand its AI-driven advertising tools. In late 2025, it introduced a model designed to improve ad performance. More recently, it launched an upgraded ad-serving system for Instagram using an Adaptive Ranking Model.
Looking ahead, analysts expect strong growth in advertising revenue. As projections suggest a significant increase in 2026, the investment in AI-driven recommendations appears closely tied to future earnings.
