Hacker News Reader

Comment by mlekoszek | Hacker News Reader

I don't understand why Facebook would want to debase the value of their content by mixing in obvious AI slop. It's a sharp contrast to their past strategy of keeping the network as 'authentic' (their words) as possible.

> AUTHENTICITY We want to make sure the content people see is authentic. We believe that authenticity creates a better environment for sharing, and that’s why we don’t want people using our services to misrepresent who they are or what they’re doing.

https://transparency.meta.com/policies/community-standards/

I'd love to hear how this happened, and what kind of rationale they could possibly have for doing it. It's very, very bizarre from an outsider's point of view.

View on HN

With no sarcasm, I truly can not understand Facebook's end game here. Why would anyone choose to spend their time interacting with fake people, pushing a corporate party line to sell you something?

Already plenty of people are tuning out of the current Facebook. How is stuffing it full of fake people going to help? Like, I can just about understand how this can goose certain short-term metrics but how would it actually make money? Fake people pushing ads on us is already a thing, that field has already been harvested, it's not like some fresh new frontier. Ads already fake this stuff, fake testimonials, fake reviews, fake web sites, fake profiles posting fake statuses for ulterior reasons. I don't see how openly hooking up LLM tech to this helps them. Covertly, I can at least cynically guess at some things, but this current approach I don't get at all.

Short of, "we've got a huge stock bump from the AI bubble, but in reality we can't actually find a way it can help us proportionally to the cost, and this is the best Hail Mary anyone has come up with to prevent an accurate revaluation of our stock."

I wonder if this is like the social media equivalent of the sims? Or a soap opera. Log in to see if jimmy works up the courage to ask out rhonda. Will ashley find out about the secret william told you? It's like a westworld where you're "safe" to interact socially in any way you feel, because they don't feel anything. As long as you click a few ads a day to keep the lights on.

People like you and me found the idea of injecting personalized ads between posts by real people disturbing. It worked fantastically for the company.

Now we think AI users subtly pushing for products such as delicious Coca Cola (TM) - refresh your day with a cool glass of Coca Cola (TM) - is a terrible idea. I expect it will again go fantastically for the company.

It just keeps getting worse and worse.

> Why would anyone choose to spend their time interacting with fake people, pushing a corporate party line to sell you something?

Really, functionally, what is the difference between any of the following other than in degree:

- Account that is nothing but AI generated content.

- Bot controlled accounts that spam messages at opportune times or can react to keywords detected in comments

- Account that belongs to an influencer and only consists of and posts influencer content.

- Account that is nothing but trolling/boosting with political memes.

- Account that belongs to a company and only posts memes and other advertisement type posts.

- Any account with activity trying to get likes, karma, etc. to gain privileges on the app or simply increase a displayed score.

All of the above are varying degrees of fake/non-authentic, and the non AI ones have been around a long, long time. These are everywhere and create engagement on multiple platforms despite people knowing about them and complaining about them. So I can understand why someone may think an AI model can possibly capture that essence and increase/lock-in engagement further. I'm pessimistic enough to think it will work if done the right way.

I was still grasping at how to phrase it, but that is definitely the sort of thing I was trying to get at, not that this is some massive change of the rules of engagement (war analogy at this point selected quite deliberately), but instead, an awful lot of money poured into plowing a field that is probably already overplowed.

And doing it openly is even weirder.

Meta (like microsoft, google, etc) have a monopoly mindset and are past thinking about what the consumer wants, they can only think about what they want. So it wasn't a case of 'would users want to follow our fake account's content?', but rather 'AI accounts will give us even more control over content, and we could use them for sponsored content and make even more money'.

My best guess: Metaverse hasn't been doing well and they want to attach the concept of AI to the company any way they can because it's the New Thing.

Makes sense. Maybe this was a "see-what-happens" experiment that caught more attention than they expected?

I can't imagine it came from Instagram having a shortage of content.

e: Like, the quality and relevance of the content is the main value proposition to users in an old school social network like Instagram. You can't really mess with that in service of some other goal without undermining the product's raison d'etre.

Generative AI seems like a great opportunity for the metaverse instead of further enshittifying their existing networks. Users could utter entire universes into existence and then explore them, this could even work for the elderly reliving some past experience (this article alludes to ensnaring older users with non-existent bot "personalities").

So why would they do this? Is it just another attempt at swaying public opinion (see: Cambridge Analytica)? Perhaps promoted messaging will come from these accounts so they can monetize their AI investment? Whatever it is, it doesn't seem like a feature users requested so it's difficult to think it benefits them.

Plausible fake Facebook friends would make it more sticky. The elderly are already subject to manipulation, are less tech savvy, etc.

Facebook, like Google, is a surveillance advertising company that tricks you into using their products so you provide targeted ad information to them. They know when or if you poop regularly l, and what brand of TP you use. Your AI friends will help them advertise better.

Almost certainly it is not doing well. Has anyone heard of anyone talking about the metaverse in real life in the last year ? Seems like one of those Covid-time fever dreams

Because all social media platforms are already flooded with obvious AI slop and they felt they were missing out?

The influx of AI slop on Facebook was apparently tied to their partner program which pays for engagement, so they were paying people to post 10,000 pictures of Shrimp Jesus. By vertically integrating the Shrimp Jesus pipeline they can save money by cutting out the middleman.

https://gizmodo.com/facebooks-twisted-incentives-created-its...

I want to say that's like a mint trying to stop counterfeiting by issuing fake bills themselves. But what am I missing here?

This surveillance advertising model wants "authentic" content for ad revenue. Natch.

It goes to assume that all statements that contradict the idea of maximum ad revenue are false. Remember "do no evil"?

Reply on news.ycombinator.com