It feels like there is a lot of AI generated content on dev.to. And the current way to prevent my feed to get cluttered by AI post that don't inter...
For further actions, you may consider blocking this person and/or reporting abuse
What would be better would be an upvote/downvote system like reddit's, eventually users with every post being downvoted just wouldn't show up in the feeds anymore.
But you know, that's if the people in charge here actually want to do something about AI posts.... which i don't think they do. [They may feel free to prove me wrong any time now]
Volunteer moderators currently have three additional moderation buttons: mark as high quality, mark as low quality, and flag for review by professional moderators.
Sometimes I'm undecided, especially those listicles that "every developer must know" are often a mix of advertising and information and they tend to be quite popular. But I downvote and flag what I am sure is spam or low-value AI slop.
here's a list of posts for you to devalue then i guess.
dev.to/member_d50fddd8
also when mods mark down a post, does it in any way affect future posts by the same author? if not, your work is kinda pointless unless you can watch every one of those accounts and flag all their posts... heh
I don't think that is a good solution because then they have to be aware of bots that up or down vote posts.
Also there are people who use AI to improve their writing style, or they are not confident enough writing in English. I don't think those people should be punished for using AI.
At the moment the filtering is in the hands of the individual users. And I think it should stay that way.
Except we are bombarded with advertising posts and spam AI "just for clicks/views" posts. So whatever they are doing is currently insufficient. One way to do this is to verify users by email, and use some tools to check user behavior to identify bots. That's a lot less work and simpler to manage, so an up/down vote system could work well. Like any other method, there will be some effort at first to make the change, but longterm it's far less work than trying to manually filter hundreds if not thousands of posts a day.
I understand your frustration.
I wasn't sure so I logged off and the only way to log in is by the usual suspects, email or social networks. So they already have that measurement in place.
What is the behaviour of a bot? Like I mentioned before people also use AI in their posts. Are we going to exclude them too?
People are promoting their stuff here, it can look spammy but I think it should be allowed.
The idea of the filtering comes from email inbox filtering. I think it works great there, so it could work on a website too.
look at this account, how is it still active and allowed to post crap, it's nothing but hyperlane ads.
dev.to/member_d50fddd8
i mean, their username alone is a red flag... come on. This is easy stuff, it's not the complex kinds of abuse, but they're not even cleaning up basic crap like this one.
I think this point is why I'm actively engaging less and less with DEV unless I'm extremely confident the post is from a person (such as this one).
AI is getting pushed so hard by the team and it's leading to the articles being so much fluff and nothing, so I only periodically check the discuss posts and ignore everything else.
I know this is the direction the industry is going and I have no problem with them getting their bag, but imo this is now a forum for AI and not a forum for developers.
Edit: Oh, I meant to mention originally that it's also quite clear that they aren't going to do anything about AI use with the recent partnership with Google AI🙃
If you read the partnership post they mention spam prevention measurements and discovering relevant content. Not sure how far along those features are.
It seems you are implying they allow AI bots? I'm not sure they are that found about the AI bots either from that partnering post.
I think there are people like you that are losing interest in the site, because they want to communicate with humans. And it is getting harder to tell a bot text from a human text.
My bot red flags are no bio, is the avatar something that you see on the site as a pattern, recently created profile and multiple posts.
AI content will get the most views, so it would be stupid of them not to allow AI posts.
I still find a lot of non AI content. And try to contribute a tiny bit.
Look at this profile and give me one good reason it's still active multiple times a week promoting hyperlane in every post. dev.to/member_d50fddd8
I don't read tutorial stuff much, most of them are AI drivel which means they're incomplete and probably have lots of innacuracies that no one addresses.
it's not even about AI for me, it's just seeing blatant advertising by a fake account like this one. dev.to/member_d50fddd8
And yea, i'm interacting less here too.
We are doing a lot to try and combat this type of low-quality content. Word filtering is an option — but AI-generated spam can be pretty smart about this sort of stuff.
I think we can probably add word filtering, but I don't think it's the absolute solution we need.
I agree it is not an absolute solution, but what is in the battle with spam content.
That's true. We're actively working on lots of stuff like this, so hang tight. We'll see if we can add this too.
Dude, yes. I could really use this too, feed just gets way too cluttered lately