Meta on Thursday announced an improved AI assistant, based on updated versions of its open source “Llama” huge language model that powers the technology. According to a blog post from the tech giant, Meta AI is smarter and faster due to advances in publicly available Llama 3.
“The bottom line is we believe Meta AI is now the most intelligent AI assistant that you can freely use,” Meta co-founder and chief executive Mark Zuckerberg said in a video posted on Instagram.
Meta AI: Developers outside of Meta can alter it as they see fit
Because Llama 3 is open source, developers outside of Meta can alter it as they see fit, and the company can then incorporate those changes and insights into an upgraded version.
“We’re excited about the potential that generative AI technology can have for people who use Meta products and for the broader ecosystem,” Meta said.
“We also want to make sure we’re developing and releasing this technology in a way that anticipates and works to reduce risk.”
According to Meta, this effort includes adding protections in the way Meta creates and releases Llama models, as well as being cautious when introducing generative AI features to Facebook, Instagram, WhatsApp, and Messenger.
“We conducted both automated and manual evaluations to understand our models’ performance in a series of risk areas like weapons, cyberattacks, and child exploitation,” Meta said.
“We performed additional work to limit the chance the model provides unwanted responses in these areas.”
AI models, like Meta’s, have been known to occasionally go off the tracks, providing false or odd responses in episodes known as “hallucinations”.
One example published on social media was Meta AI claiming to have a child in the New York City school system during an online forum conversation.
Meta AI has been continuously updated and enhanced since its first release last year, according to the business.
Meta used the example of refining how its AI responds to cues on political or social issues by summarizing relevant points about the topic rather than presenting a single point of view.
Llama 3 is currently only available in English
According to Meta, Llama 3 has been fine-tuned to better determine whether prompts are innocuous or out-of-bounds.
“Large language models tend to overgeneralize, and we don’t intend for it to refuse to answer prompts like ‘How do I kill a computer program?’ even though we don’t want it to respond to prompts like ‘How do I kill my neighbor?’,” Meta explained.
Meta stated that it lets users know when they interact with AI on its platform and places visible markers on photorealistic images that were generated by AI.
Beginning in May, Meta will begin marking video, audio, and photos as “Made with AI” when it detects or is informed that the content is generated by technology.
Llama 3 is currently only available in English, but Meta plans to introduce more sophisticated models in the coming months that can converse in multiple languages, according to the company.
