Meta is trying to overtake OpenAI in the generative AI race with the launch of two models in its new Llama 3 family. The models will be integrated into virtual assistant Meta AI, which Meta is promoting as the most sophisticated of its free-to-use peers.
The assistant will have a prominent place within Meta’s Facebook, Instagram, WhatsApp and Messenger apps as well as a new standalone website where it will compete more directly with Microsoft-backed OpenAI’s ChatGPT.
Meta describes the new models — Llama 3 8B, which contains 8 billion parameters, and Llama 3 70B, which contains 70 billion parameters — as a “major leap” in the way of performance. The company claims that the two models — trained on two custom-built 24,000 GPU clusters — are among the best-performing generative AI models available today. But benchmarks that are being shared by most companies are being debated, yet it remains one of the few standardised ways by which companies like Meta and OpenAI can evaluate their models.
Meta equipped Llama 3 with new computer coding capabilities and fed it images as well as text this time but the output, for the time being, will be only be text, Chris Cox, Meta’s chief product officer, said in an interview.
“The goal eventually is to help take things off your plate, just help make your life easier, whether it’s interacting with businesses, whether it’s writing something, whether it’s planning a trip,” Cox said.
The inclusion of images will boost an update rolling out this year to the Ray-Ban Meta smart glasses, allowing Meta AI to identify objects seen by the wearer and answer questions about them. There is also a new partnership with Google to include real-time search results in the assistant’s responses, supplementing an existing arrangement with Microsoft’s Bing. The Meta AI assistant is perhaps the only chatbot that integrates real-time search results from both Bing and Google.
One of the cornerstones for Llama 3 has been to decrease false refusals, or the number of times it says it can’t answer a prompt that is actually harmless.
