- cross-posted to:
- ghazi@lemmy.blahaj.zone
- cross-posted to:
- ghazi@lemmy.blahaj.zone
You must log in or # to comment.
How do we keep the bots out of Lemmy? Eventually all the AI rot will spread to the fediverse.
Defederate/ban them, defederate instances that don’t adequately stop them from their instances.
That might work for now, but AI bots can easily create more content faster than human moderators can go through. We will need mechanisms to prove humanness.
Another strategy might be to demand a minimum content quality. Whether trash comes from humans or bots, it’s still trash.
They rarely get past the signup form, it’s trivial to spot them.
That’s how you dismantle the current network in it’s entirety
Realistically, you don’t. You can’t really.
The only way would be to verify each account somehow.
Yea, it will be a hard problem to solve. There’s no good solutions right now. Hopefully someone with the creativity and skill will come up with something.
It becomes a pain vs gain problem. How hard do you make it, also balancing the inconvenience.
You could easily force users to enter a one time code via email every 3 months (or more or less time). This would be hard to automate and if you changed it up even more so.
Or maybe enforce login with a hardware token like Yubikey or Nitrokey. I don’t think you could automate that(?)
Yeah but how do you get it to someone.
Pain vs gain.
I’m not sure I understand. User would have to buy their own keys and set it up.
Are AI bots really the true problem, or is the problem product promotion? Humans can do that too.
Yea, but AI bots do it at an unprecedented scale. The price of generating bullshit is practically 0. There is already cheap AI content every where and people are already getting tired of it. It’s true that human actors has done something similar for a long time, but now AI will make the process so easy that AI junk will be everywhere. Look at Google search results, it’s already filled with blogspam generated by AI. Blogspam was a problem before, but now it’s insufferable. Cheap midjourney “art” is showing up in search results even when you search for real classic artist, you get these ugly AI copies. Generative AI is the ultimate bullshit and spam machine.
Thank you for the insightful response!
Bots aren’t the problem. You mistake the tool for its operator by cursing a bot. If I had a specific purpose for a bot, I’d use one. The difference between my using a bot and a clickfarmer using a bot is that I wouldn’t treat Lemmy instances as a resource to be exploited (and ultimately ruined). Such an attitude would inform the how and the what behind my bot use.
The problem as always comes back to bad incentives, bad design and bad behaviour.
We are a community. The community will find a way as we can do anything. We have faith of the heart and everything.
That sounds more like a prayer than a plan.
I can see the measures we currently take for ads (large user-maintained subscribable blocklists) being part of the solution. That, and some form of privacy-respecting POW scheme.
And it couldn’t have happened to a nicer website.
It was like that before I left
I’m not a regular there any more, but I feel like there’s more “natural” spam posts nowadays. They’re not natural enough that they haven’t prompted me to check the account’s post history (all mention product x, of course).
Edit: e.g. this one
If this is an example of the AI, it’s not very good.
LLMs don’t generate that sort of templated response for conversation (although they do for software code).
It’s more likely a traditional ad, using some classical algorithms perhaps.
