Meta has been offering an AI summary of Facebook posts to a few users. The social media giant’s in-house Generative Artificial Intelligence (Gen AI) seems to be generating these summations of what people are saying or commenting.
Meta AI generating bizarre summaries for Facebook posts
Meta has begun offering or suggesting a summation of what people are saying about a Facebook post. The company’s Gen AI appears to be offering a gist, tone, mood, and even inclination, of the content that users post as a response.
Instead of showing me the comments on my congressman’s Facebook post it gave me an AI summary of the comments.
byu/Striking_Sea_129 inmildlyinfuriating
Strangely, several comment sections on Facebook default to an AI summary instead of showing the comments. It is not clear why Meta is confident that Facebook users want a summary generated by a Gen AI, instead of reading the comments themselves.
Several social media users have indicated that the AI summaries are popping up on many of their posts. And as expected, they are quite bizarre. These AI summaries appear quite similar to the ones Amazon has been offering to buyers.
Facebook AI literally makes a summary of the comments for you. What a useless feature. That’s not how I like to be nosy. pic.twitter.com/AVTLFjRymX
— 🍑🥤 (@peachmilfshake) May 31, 2024
Amazon’s Gen AI seems to be quite accurate in offering a common set of opinions. Meta’s Gen AI, on the other hand, seems to be merely including some comments, and ignoring others.
How does Meta decide what to include in the condensed reports?
It is not clear how is Meta sifting through hundreds or thousands of comments. However, according to The Verge, these Facebook AI summaries are quite random. In other words, the AI appears to be randomly choosing comments to form a quick summary.
It is amply clear that Meta is testing AI summaries for Facebook posts. Moreover, comments on social media posts have an insanely wide range or spectrum. Hence, the Gen AI generating these summations could easily get overwhelmed.
1. I’m legit shocked by the design of @Meta‘s new notification informing us they want to use the content we post to train their AI models. It’s intentionally designed to be highly awkward in order to minimise the number of users who will object to it. Let me break it down. pic.twitter.com/rhKNFt7CEu
— Tantacrul (@Tantacrul) May 26, 2024
Meta could finetune its Gen AI by allowing Facebook users to suggest a better summary, or rate them on multiple parameters, including accuracy. These AI summaries have started to randomly appear with seemingly no way for users to voice their opinions about them.
Another major potential issue with Facebook AI summaries could be privacy concerns. Meta is feeding user comments into its AI system to generate them.
friends, i followed all these steps to the letter and meta still rejected my request because they claimed they couldn’t find examples of their generative ai model using my personal information *yet*, so they had no reason to opt me out of a policy i explicitly denied consent for. https://t.co/JK50patGJw
— mary 🌱 (@merryfight) May 28, 2024
Meta has already started informing Facebook and Instagram users in the European Union and the UK about the same. The company does allow users to object to this process and deny the use of their data to train the AI. However, the process to stop Meta from using data to train its AI isn’t straightforward.
2024-06-01 15:06:26