Meta’s GenAI research, from Emu to Audiobox
Meta’s AI research director Devi Parikh recently sat down with Meta engineer Pascal Hartig on the Meta Tech Podcast to discuss the company’s trajectory in generative AI — where it’s been, where it’s heading, and the papers driving the field forward.
The conversation touched on several of Meta’s recent model releases, giving engineers a look at the breadth of work underway across modalities.
Key models and tools discussed
- Audiobox: A foundational model for generating audio and soundscapes from natural language prompts.
- Emu: Meta’s first foundational model for image generation, marking an early step in its visual GenAI efforts.
- Purple Llama: A suite of tools aimed at helping developers deploy GenAI models safely and responsibly.
The episode is available now on Spotify, PocketCasts, Apple Podcasts, and Google Podcasts. The Meta Tech Podcast is produced by Meta and covers engineering work spanning low-level frameworks to end-user features.
Feedback on the show can be sent via Instagram, Threads, or X, and those interested in AI roles at Meta can find openings on the Meta Careers page.
