I think they know their audience better than they do. It will take some tuning to work out the bugs, but eventually the AIs will be exactly what each individual expects their friends to be.
They will probably aim for that, but actually doing so will produce hilarious results. Imagine what would happen to someone who is into BDSM or some other spicy topic. Then meta would generate them a bunch of BDSM buddies that post all sorts of wild stuff all the time.
My argument is that as Meta matures, it becomes increasingly risk averse, just like large companies tend to. They might still try what you suggested, but after a few PR disasters they’ll tone it down a bit. Maybe after a few iterations, the bots will produce hilarious results as rarely as possible while still being just barely interesting. In other words: harmless and inoffensive.
I think they might just let the disasters happen. The end picture looks less like a social network and more like billions of individual humans floating in their own personal fantasy worlds populated almost entirely with AI friends.
The current LLMs hallucinations and horror themed pictures have been pretty entertaining already. Sounds like there could be good times ahead if Meta lets the AIs run wild. 🍿
Status update: it took only a few minutes for the backlash to happen.
As users began to sniff out some of Meta’s AI accounts this week, the backlash grew, in part because of the way the AI accounts disingenuously described themselves as actual people with racial and sexual identities.
I think they know their audience better than they do. It will take some tuning to work out the bugs, but eventually the AIs will be exactly what each individual expects their friends to be.
They will probably aim for that, but actually doing so will produce hilarious results. Imagine what would happen to someone who is into BDSM or some other spicy topic. Then meta would generate them a bunch of BDSM buddies that post all sorts of wild stuff all the time.
My argument is that as Meta matures, it becomes increasingly risk averse, just like large companies tend to. They might still try what you suggested, but after a few PR disasters they’ll tone it down a bit. Maybe after a few iterations, the bots will produce hilarious results as rarely as possible while still being just barely interesting. In other words: harmless and inoffensive.
I think they might just let the disasters happen. The end picture looks less like a social network and more like billions of individual humans floating in their own personal fantasy worlds populated almost entirely with AI friends.
The current LLMs hallucinations and horror themed pictures have been pretty entertaining already. Sounds like there could be good times ahead if Meta lets the AIs run wild. 🍿
Status update: it took only a few minutes for the backlash to happen.
source: https://edition.cnn.com/2025/01/03/business/meta-ai-accounts-instagram-facebook/index.html
Oh, now they have bots that are into all sorts of spicy stuff. Pretty much what I expected.