# Meta AI 2024 Meta AI is the artificial intelligence research division of Meta Platforms, founded in 2013 as Facebook AI Research (FAIR). It develops AI technologies including the Llama large language models, the Meta AI virtual assistant, and custom AI hardware. Meta AI is a research division of Meta Platforms (formerly Facebook, Inc.) that develops artificial intelligence and augmented reality technologies. As of 2025, it maintains workspaces in Menlo Park, London, New York City, Paris, Seattle, Pittsburgh, Tel Aviv, and Montreal. The division is known for its open-source contributions to [machine-learning](https://www.wikiprompt.org/wiki/machine-learning) and [deep-learning](https://www.wikiprompt.org/wiki/deep-learning), including the PyTorch framework and the Llama family of [large-language-model](https://www.wikiprompt.org/wiki/large-language-model)s, as well as for the Meta AI virtual assistant integrated into Meta's social networking products. Meta AI traces its origins to 2013, when Facebook established Facebook AI Research (FAIR) under the direction of Yann LeCun. Over the following decade, FAIR grew into a global research organization, publishing influential work in areas such as [generative-ai](https://www.wikiprompt.org/wiki/generative-ai), computer vision, and natural language processing. In 2025, Meta reorganized its AI efforts under Meta Superintelligence Labs, which consolidated FAIR, large language model development teams, and other AI research and product groups. ## History Meta AI was founded in 2013 as Facebook AI Research (FAIR). The lab was directed by Yann LeCun until 2018, when Jérôme Pesenti succeeded him. Under LeCun, FAIR became known for pioneering research in [self-supervised-learning](https://www.wikiprompt.org/wiki/self-supervised-learning) and [generative-adversarial-network](https://www.wikiprompt.org/wiki/generative-adversarial-network)s, as well as for contributions to document classification, machine translation, and [computer-vision](https://www.wikiprompt.org/wiki/computer-vision). In 2016, FAIR released fastText, an open-source library for efficient text classification and learning word representations. FastText introduced methods based on character n-grams, which improved performance on morphologically rich languages and enabled faster training on standard hardware. In 2017, FAIR released PyTorch, an open-source [machine-learning](https://www.wikiprompt.org/wiki/machine-learning) framework built on the Torch library. PyTorch's dynamic computation graph and Python-first design made it widely adopted in both academic research and industry, becoming one of the dominant frameworks for [deep-learning](https://www.wikiprompt.org/wiki/deep-learning) alongside TensorFlow. Following the 2021 rebranding of Facebook, Inc. as Meta Platforms, FAIR continued as Meta's Fundamental AI Research team. In 2023, Meta introduced the Meta AI assistant, based on the company's Llama models, and integrated it into its social networking products. In 2025, Meta reorganized its artificial intelligence efforts under Meta Superintelligence Labs, which brought together FAIR, large language model development teams, and other AI research and product groups. ## Virtual assistant Meta AI is also the name of the virtual assistant developed by the team. The assistant is integrated as a chatbot into Meta's social networking products, including Facebook, Instagram, and WhatsApp, and is also available as a stand-alone app. Optional paid subscriptions provide higher usage limits for the assistant's capabilities. The virtual assistant was pre-installed on the second generation of Ray-Ban Meta smartglasses, and can incorporate inputs from the glasses' cameras after an update. It is also available on Quest 2 and newer head-mounted displays (HMDs). Since May 2024, the chatbot has summarized news from various outlets without linking directly to original articles, including in Canada, where news links are banned on its platforms. This use of news content without compensation and attribution has raised ethical and legal concerns, especially as Meta continues to reduce news visibility on its platforms. ## Research Meta AI's research spans multiple areas of artificial intelligence, including [natural-language-processing](https://www.wikiprompt.org/wiki/natural-language-processing), computer vision, and AI infrastructure. The division publishes extensively in academic venues and releases many of its tools and models as open-source software. ### Natural language processing One of Meta AI's research areas is natural language processing, including machine translation, natural language generation, and question answering. In 2023, the Meta AI team released a set of natural language processing models named Seamless, designed for enabling real-time voice translation. In 2024, a team including researchers from Meta AI published a paper about developing machine translation for languages not typically supported by machine translation. An editorial article in the journal Nature commented on the paper, emphasizing the importance of involving people who specialize in languages in the development and application of machine learning technology. #### Galactica Galactica is a large language model (LLM) designed for generating scientific text. It was available for three days from 15 November 2022, before being withdrawn for generating racist and inaccurate content. The model was intended to help scientists navigate the vast literature, but its tendency to produce plausible-sounding but incorrect output led to public criticism and its rapid removal. #### Llama Llama is an LLM released in February 2023. As of January 2026, the most recent release is the Llama 4. The Llama family has been released under open-source licenses, making it a popular choice for researchers and startups seeking alternatives to proprietary models from [openai](https://www.wikiprompt.org/wiki/openai) and [google-deepmind](https://www.wikiprompt.org/wiki/google-deepmind). Llama models have been used for a wide range of applications, from chatbots to code generation, and have spurred a broader ecosystem of fine-tuned variants. ### Muse In April 2026, Meta Superintelligence Labs introduced Muse Spark, the first model in the Muse family. Muse Spark is a multimodal reasoning model designed for tasks including tool use, visual reasoning, and multi-agent workflows. Meta subsequently expanded the family with specialized models including Muse Image and Muse Video, which focus on image and video generation and understanding. ### AI infrastructure Meta has developed large-scale computing infrastructure for artificial intelligence research and deployment. In 2022, it introduced the AI Research SuperCluster (RSC), a GPU-based supercomputer designed to train large models in areas including natural language processing and computer vision. RSC was built with thousands of Nvidia GPUs and a high-bandwidth interconnect to support training of models with hundreds of billions of parameters. Meta also develops the Meta Training and Inference Accelerator (MTIA) family of custom AI accelerators for its internal workloads, while continuing to use large clusters of Nvidia GPUs. The MTIA chips are designed to optimize inference and training for Meta's specific recommendation and ranking models, reducing reliance on third-party hardware. ## Controversy The French media outlet Mediapart reported that in 2022, Facebook's parent company illegally used works accumulated by the pirate site LibGen to train its artificial intelligence. The report alleged that Meta used a dataset derived from LibGen, a repository of copyrighted books and academic papers, without authorization. Meta did not publicly confirm or deny the specific details, but the incident highlighted ongoing debates about the use of copyrighted material in training [large-language-model](https://www.wikiprompt.org/wiki/large-language-model)s. ## See also - [artificial-intelligence](https://www.wikiprompt.org/wiki/artificial-intelligence) - [generative-ai](https://www.wikiprompt.org/wiki/generative-ai) - [large-language-model](https://www.wikiprompt.org/wiki/large-language-model) - [machine-learning](https://www.wikiprompt.org/wiki/machine-learning) ## References (Source facts provided; no external links.) ## External links (No external URLs per instructions.) --- Source: https://www.wikiprompt.org/wiki/meta-ai-2024 License: CC BY-SA 4.0 (https://creativecommons.org/licenses/by-sa/4.0/) Last updated: 2026-09-13T03:55:42.522296+00:00