Meta AI is a company owned by Meta (formerly Facebook) that develops artificial intelligence and augmented and virtual reality technologies.
It was previously known as Facebook AI Research (FAIR) before being renamed following Meta’s rebranding in 2021.
Meta AI focuses on academic research to advance the field of AI, rather than on practical applications for Meta’s products.
Its initial work included research in areas like learning-model enabled memory networks, self-supervised learning, generative adversarial networks, and natural language processing.
In 2022, Meta AI released a large language model called Galactica designed for generating scientific text, but withdrew it shortly after due to issues with offensiveness and inaccuracy.
More recently in 2023, Meta AI launched LLaMA (Large Language Model Meta AI), a large language model ranging from 7 to 65 billion parameters.
Meta AI is also developing hardware for AI training and inference, including the MTIA v1 accelerator chip. It offers AI-powered features across Meta’s family of apps, such as the ability to generate images from text descriptions, answer questions, and provide recommendations.
However, Meta’s use of AI to summarize news articles without linking to original sources has raised some ethical concerns.