Comment by everdrive
2 days ago
I'm a little bit confused about what's going on here. Is this nothing more than an LLM-generated summary of her post? She shows the metadata but also shows it coming up in the post. I don't use any of these apps so I'm not really sure what a normal user would have seen. ie, would that text have been appended visibly to her post, making users think she wrote that, but also have been in tags which would have optimized for search engines?
Either way, I don't know what to tell people. Social media exists to take advantage of you. If you use it, your choices are "takes more advantage" vs. "takes less advantage," but that's as good as gets.
It looks like it's a third-party UI, her Mastodon client, using the description metadata in a way that kind of makes it look like that metadata is part of the post.
Auto-generating said description tag in the first person is a bit of a weird product decision - probably a bad one that upsets users more than it's useful - but the presentation layer isn't owned by Meta here.
Thanks for the explanation, that makes a lot of sense. I'll bet that when it's not a sensitive topic, this totally goes unnoticed by a lot of users. Frustratingly, I would imagine that the response from most people would just be that the LLM summarizations / metadata tagging should be censored in "sensitive cases," but will otherwise be accepted by the user base.
Author posted to Instagram > Author shared the Instagram link on Mastodon > Mastodon mobile app unfurled the link into a preview > app concatenated mystery text from a hidden metadata field in the Instagram page > turns out Meta's LLM wrote first-person inspiration slop in the "I" voice for SEO > Author feels impersonated