Meta uses AI to summarize Facebook comments

Have you seen Meta’s artificial intelligence (AI) pop up in Facebook comments lately, summarizing what people are saying? With these meetingsArguments are often “tense” on Facebook, sometimes AI summarizes absurd things. (This is not the first time Meta’s AI has appeared in the comments section. Previously, 404 Media discovered an AI pretending to be a parent in a Facebook group).

I checked the comments on Facebook and saw the summary AI appear on many posts, including the “javelin” comments. In a post about the store’s closure, AI summarized that “Some commenters suggested that the store closed because of ‘迎合 wokeness’ (a tendency to focus on socio-political issues) or product selection. poor products, while others blame the rise of online shopping.”

Another post from Vice about Mexican street wrestlers was summarized by AI saying some people were “less impressed” with the performance and called it “stupid begging”. The AI ​​also understood some of people’s light-hearted jokes about feral cat sightings in Florida. “Some people enjoyed the sighting, with one commenting hoping the wildcat remembered to bring sunscreen.”

Currently, how Meta chooses which posts to display comment summaries is unclear, and the company has not spoken out on the matter.

Anyway, AI summaries aren’t really that useful (unless you like vague ideas about what people say). However, it can help you identify posts whose comment sections are too “toxic” to scroll down to see.

AI summaries also raise privacy concerns, as Meta is feeding user comments into the AI ​​system that generates them. Over the past week or so, many Facebook and Instagram users in Europe and the UK received notifications that Meta would train its AI on their content (required by data protection laws in both regions). Meta must disclose this information). Although Meta allows these users to object to the use of their data for AI training, the process is not simple and the company has denied some users’ requests.

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

In the US, Meta’s privacy policy page says the company uses “information shared across Meta Products and services” to train its AI, including posts, photos and captions. Meta allows you to submit requests to correct or delete personal information used to train their AI models, but it only applies to information from third parties. Everything else seems to be used by Meta.