Meta is currently working on a fundamental shift in how billions of users interact with its multiple social platforms. Recent reports have revealed that the technology giant has formed a secret team comprising top artificial intelligence researchers, with the primary goal of carrying out a comprehensive overhaul of all recommendation algorithms that determine the nature of content shown to users on Facebook, Instagram, and Threads.
The inception and focus of the new research unit
The new unit, named “MRS Research,” operates within Meta’s recommendation systems division. Although the unit was formed in October 2025, it has operated out of the spotlight and in complete secrecy throughout the past period. The unit is distinguished by its direct and close connection to the company’s advertising division, the primary driver behind Meta generating massive revenues totaling $201 billion in 2025. Through this collaboration, the unit aims to improve both organic content and ad delivery methods in users’ main feeds, ensuring they stay on the platforms for longer periods.
Veteran leadership from TikTok
In a move reflecting Meta’s seriousness in competing with fast-growing apps, MRS Research is led by technical expert Yang Song, who joined Meta in November 2025 as vice president of recommendation research. Song is a major asset for Meta, having previously served as head of recommendation algorithms and user growth at TikTok, where he managed a massive team of over 400 engineers and researchers specializing in machine learning. Song’s experience at TikTok, renowned for its highly engaging and accurate algorithms, is considered a crucial element in Meta’s future vision.
Song explained in a LinkedIn post last December that the new lab’s mission centers on bringing about a qualitative shift in recommendation systems. He noted that this will be achieved by introducing an entirely new model relying on large language models, leveraging their massive capabilities in absorbing global knowledge, understanding long and complex contexts, and utilizing advanced reasoning and logical analysis capabilities to accurately predict user preferences.
Attracting top AI minds
To achieve these ambitious goals, Meta has not only appointed strong leadership but has also launched an intensive hiring campaign to poach talent from competing tech companies. According to LinkedIn profiles and press reports, the team has successfully brought in a group of prominent artificial intelligence researchers. Among them are Lehong Li, who previously worked at Amazon, Xiaolong Wang, a former researcher at OpenAI, and Fei Sha, who moved from Google. This exceptional gathering of expertise underscores the scale of investment Meta is pouring into this vital project.
Separating applied AI from superintelligence
It is important to distinguish between the MRS Research unit and Meta’s other AI efforts. A company spokesperson clarified that the new unit is completely separate from “Meta Superintelligence Labs,” the more prominent unit led by Alexander Wang, former CEO of Scale AI. While Superintelligence Labs focuses on developing general and advanced AI models to reach new horizons, the MRS Research unit focuses narrowly and intensively on applied sciences related to recommendation systems—specifically understanding content nature, analyzing user behavior, data retrieval processes, and content ranking across Meta’s family of apps.
Financial impact and advertising development
This massive investment comes at a time when Meta increasingly relies on AI-driven ad optimization technologies. In late 2025, the company launched a new model it confirmed significantly boosts ad effectiveness. Recently, Meta unveiled an improved ad delivery process on Instagram relying on advanced technology to rank content and adapt to changes in user behavior. Financial institutions expect a positive payback from these developments; Morgan Stanley issued a forecast on March 30 indicating that Meta’s ad revenues will see significant growth of 28 percent during 2026, confirming that investment in smart algorithms is the shortest path to boosting profits.
Frequently Asked Questions
What is the “MRS Research” unit recently formed by Meta?
It is a secret research unit formed by Meta in October 2025, aiming to carry out comprehensive development on recommendation algorithms that determine the content users see on Facebook, Instagram, and Threads, using large language models.
Who is leading this new unit?
The unit is led by expert Yang Song, who joined Meta as vice president of recommendation research in November 2025. Song previously served as head of recommendation algorithms and user growth at TikTok, where he managed a team of over 400 specialists.
What is the difference between the Recommendation Research Unit and Meta Superintelligence Labs?
The MRS Recommendation Research Unit focuses on the practical application of AI to improve content display, user understanding, and post ranking. Meanwhile, “Meta Superintelligence Labs,” led by Alexander Wang, focuses on developing general AI models with exceptional and comprehensive capabilities.
How will these technical developments affect Meta’s financial standing?
Optimized recommendation algorithms are expected to drive higher user engagement and improve the targeting accuracy of ads. Based on these developments, Morgan Stanley projected that Meta’s ad revenues will grow by 28 percent in 2026.
