Hacker News new | ask | show | jobs
by r3trohack3r 1111 days ago
I don’t think social media is going to survive AI, at least not in its current form.

It’s not just going to take a blow from the AI content production that’s on the horizon but also AI engagement.

Pretty much all of the signals social media platforms use to automate curation of content are about to turn into noise against the backdrop of nearly every participant in the social network being both incentivized and capable of running a Sybil attack with a seemingly infinite team of AI content producers and profiles capable of driving engagement.

3 comments

I think the next stage is decentralized social media. Something like nostr (1) where there’s no centralized entity determining the algorithm to boost. It’s up to the individual to follow users.

Perhaps the next challenge would be human verification, even with this protocol we’d need something to index public people by to handle discovery.

Even before LLM’s became as mainstream as they are, most social media platforms were riddled with spam: affiliate marketing, drop shipping crap, and people who are running some sort of con.

1 - https://github.com/nostr-protocol/nostr already has 8k stars on github

I think this decentralized social media future is already here, and it's private discord servers
The past is the future, and it was called IRC
GitHub stars? You mean the stars you can purchase through third party bot services? (As shown here recently on HN!)
The next stage is requiring a phone number to sign up and having no privacy online.
The AI flood coming is the authoritarians wet dream for multiple reasons.
It isn't exactly a new theory that a significant amount of traffic on social media is driven by AI. https://en.wikipedia.org/wiki/Dead_Internet_theory

I'm a little surprised how long Twitter has managed to last in it's current form. I don't know any real life people, besides celebrities, who actually actively use twitter. I never understood how it is able to sustain itself. But then again... tabloid magazines are still sold in grocery stores even though I've never seen anybody ever purchase one... Operating costs must be low.

Now that AI is widely available, I think it will be in social media platforms' interest to develop verification methods that make sure that real people are using the platform and not AI.

Twitter already feels passed the tipping point, although that's purely just based on how it feels. On that note, there should be a name for when you start to mistake genuine human activity for being the product of AI. And obviously Philip K. Dick was onto this sort of thing decades ago...
Twitter is uniquely toxic among all the other social networks. When the human-generated activity starts resembling a Markov chain of bile, it really doesn’t matter if AI takes over. I’m some sense, it already has: the platform is a system impressing itself on users who propagate its values.

A system is what a system does.