We just don’t smell the rot yet. It’s not about the actual body (website) and the name. It is about the spirit, the personality. And that has been compromised and killed off because of factors one could call enshittification or whatever. Digg and Tumblr still exist, at least in name, but not in spirit. I think Reddit has past that threshold. It will be a slow death and I am not sure the fediverse will fill the void it leaves behind. But I do think and believe communities will never sustain on venture capital.
I’m just waiting to see if any of the big AI companies they claim are going to pay actually do. I’m willing to bet money that API usage levels dropped off a cliff as soon as the new pricing took effect, and that there will be no big payouts from Microsoft etc
You know… I don’t understand why AI would need access to an API. Can’t they just crawl the web? HTML5 was designed with AI in mind iirc. But I’m no expert so I’ll probably talk bs here.
They can get all the data they want without the API, and I think they will from here on out. But if a site happens to provide data in a convenient API format for free, they might use it. Or they may not, I don’t know what evidence Reddit used to decide that the LLMs training on Reddit data was doing it via the API. It’s possible that all Reddit ever intended to do was kill third party apps (and it seems like they’re still settling for killing most, and maiming the stragglers).
I think it’s only a matter of time before Reddit sets up rate limits and blocks unregistered views like Twitter in another misguided attempt to stop AI from pulling in massive amounts of user data.
Problem is that there’s nothing stopping an AI company from just creating thousands of dummy accounts to get as much data as they need before each hits the daily rate limit anyways.
Short of locking all data away that is older than a specific time, it’s a losing battle no matter how you look at it. And going that far is sure to doom the service anyways.
I think Reddit would like to keep people clicking through from Google searches, which, apparently, Twitter doesn’t, but I suppose they might. And yeah, I think bots are basically inevitable, one way or another.
On the other hand, with increasing amounts of bot-generated content on Reddit, it might not be long before no one wants to train LLMs on Reddit anymore, lol.
The big AI companies will almost certainly not pay the usage fees as they are for everyone else. What this might accomplish (and might be the goal) is to bring them to the negotiation table for special deals, lump sum payments etc.