Meta’s Meta Connect event last week showcased new AI and mixed reality innovations.
Tech giant Meta announced that its new chatbot Meta AI, powered by the open-source big language model Llama 2, was trained with public Instagram and Facebook posts.
Clegg, who made a statement on the issue after tech companies were criticized for using unauthorized data from the internet to train AI tools, said that the vast majority of the data used for training were public posts.
Meta said it did not include private messages to family and friends in the training because of the importance of user privacy, nor did it use private chats in messaging services.
“We also tried to exclude datasets that are dominated by personal information,” added Nick Clegg, Meta’s Head of Global Affairs.
Another notable feature of Meta AI is that you can start messaging on WhatsApp or Instagram with a single click while in any Meta app.
This eliminates the need for users to constantly log in between different apps, making their communication easier and smoother.