Introducing Meta AI: Meta’s Answer To ChatGPT And More
Meet Meta AI, the new virtual assistant powered by the Llama 2 language model. Like OpenAI’s ChatGPT, this virtual assistant answers questions from Microsoft’s Bing search engine and generates images from text commands using the Emu image generator, trained on over a billion pairs of texts and photos, including content from Instagram and Facebook.
Meta AI will be initially available to a select few in the United States on different platforms. Aside from assisting with tasks such as travel planning, it can be added to group chats. It will also work with voice commands through new smart glasses that Meta will release next month.
Meta’s announcement of Meta AI and its image generator puts it in competition with OpenAI’s ChatGPT and recent text-to-image generator releases. While Meta is late to the game compared to Microsoft, Alphabet, and Amazon in launching or upgrading chatbots, this move is part of Meta’s effort to expand its AI capabilities.
Additionally, Meta introduced Backdrop and Restyle, generative photo editing tools that use AI. Backdrop swaps backgrounds based on text prompts, while Restyle uses generative AI to create various effects around a person, such as surrounding them with puppies. These tools will be available to Instagram users in the near future, and images created with them will feature a watermark to distinguish them from real photos.
Meta also unveiled various chatbots based on celebrities, including Paris Hilton, Tom Brady, and Snoop Dogg, available in beta on Instagram, Facebook Messenger, and WhatsApp. Meta plans to make the tools used to create these celebrity chatbots available to users and businesses in the future.
Unlike Google and OpenAI, which keep their latest AI models proprietary, Meta is committed to open-sourced machine-learning models, such as Llama 2. This approach encourages outside developers to adapt and improve its models, benefiting its apps and services, like the code-generating Llama variant released last month.