Meta's New AI Assistant: Trained with Your Instagram and Facebook Posts

After Meta’s Data Breaching Past Do We Really Want Them Snooping On Our Content?

Meta, the company that used to be called Facebook, has just told everyone that they used public posts from Instagram and Facebook to teach their new AI creation. This new AI tool can do all sorts of things, like answer questions, help plan trips, and even make pictures from what you write. Pretty much like all other AI tools but with the meta team behind.

Meta are saying they didn’t use private stuff, so it’s not like your secret chats were used. But can we really rely on this kind of information or is it going to be another scandal years down the line.

How They Trained Meta’s AI with Public Posts

Nick Clegg, who works as Meta’s President of Global Affairs, says that most of the things that taught the AI helper came from things anyone can see on Instagram and Facebook. He also said they tried not to use posts that had too much private info in them. What we would love to know is how did the seperate the posts from what’s private and what isn’t? did it use AI to do that too, because we all know how AI loves to hallucinate.

They used text and pictures from public posts to teach Emu, the AI’s picture-making part. The talking part, which is like a chat, used Llama 2 to learn from stuff online. They mixed in some other information too, like things you can find online, but they made sure to say it’s from the internet. Just think of it like ChatGPT but Metas version. It’s supposed to be more light weight.

Meta’s AI buddy uses really smart technology from a company called Qualcomm. It can do cool things like showing pictures, taking pictures, talking, and even looking up answers on the internet. Meta’s also planning to bring in 28 AI characters to talk to you on their apps. Some of them will look like famous people, like Snoop Dogg or Kendall Jenner! We think this a bit silly and so do other celebrities.

Why People Are Worried About Meta Using Public Posts

Meta getting its AI information from public posts has made some people worried. They’re thinking about how this could be a problem:

Privacy: Even though Meta says they only used public posts, some people might not know that their posts are public. Or they might not want their posts used for training a computer. Some people could’ve changed or deleted their posts after Meta collected them. That might make people feel weird knowing a big company used their pictures and messages for its own stuff.

Fair Use: Meta says they used public posts because it’s kind of like borrowing a book from a library to learn stuff. They say it’s okay because of something called fair use, which means you can use things like books or movies to learn or teach. But, some smart people think this isn’t the same. They say it’s not fair because Meta didn’t make the posts better or teach anything new. Plus, fair use is like a puzzle, and it’s different for every case.

Money: Some people who posted stuff on Instagram and Facebook might think they should get something for helping Meta’s AI learn. They might say their posts are like a treasure, and Meta just took it without asking or saying thanks. They could want more info about how their stuff is used and maybe even some money for it.

In the future we might even have AI influencers

In the End

Meta saying they used public posts from Instagram and Facebook for their AI has started a big conversation. People are talking about their privacy, fair use, and getting something in return for their stuff. As AI becomes even more important, these are going to be even bigger questions in the future. We have already seen what AI can do when it comes to creating images and even cloning creative writers. Do we really want AI to steal our posts too?