Mine got flagged twice in one day.
Both essays generated more engagement than anything else I've published here. One had a five-exchange technical comment thread building a production architecture on top of it. The other had an AI podcast episode generated from it before it even published.
The founder of DEV.to liked one of them before the bot flagged it.
I'm not here to argue about the policy. The guidelines are reasonable. AI disclosure is fair. I added the disclaimers and moved on.
But I want to know if this is just me.
If Sloan has flagged you recently, drop a comment. Three things I'm curious about:
- What kind of article was it — tutorial, opinion, essay?
- Did you use AI assistance, and if so how much?
- Did you think the flag was fair?
I want to know if there's a pattern or if I got unlucky twice in one day.
Top comments (27)
Hey Daniel. Here to break the news for you.
Not sure if other Moderators are doing this, but the 99% of the majority of Sloan was from me.
This was hard to tell you for many reasons. After releasing an article about this issue and taking the time to decide on what to do, I decided to take initiative and go around to find articles that are AI-assisted and enforcing the DEV rule on the use of AI. Since moderation is kind of inactive (in my view), I decide to take action appropriately as a moderator because I care about the community and its content.
Of course, all Sloan message are not about AI, I sent a message to @hemapriya_kanagala on the formatting issue. I use Sloan appropriately to ensure the author uses best practices and improve on their content, not to use it as power.
I deeply care about the community and its content being produce. However, I notice that some of the rules on DEV are not enforced and when I started seeing an increase of bot posts/comments/engagement, it is best practice as a moderator to notify authors to use best practice on writing articles. I don't want the platform to be TOO conformable to the point where they use AI irresponsibly and can view negatively and discourage new users who are trying to grow as a developer and connect with actual people.
At first, it was difficult to send a Sloan message to you because we known each other in the platform since Richard was here. However, I don't want to play favorites and decide on sending it to you. Not only to encourage you to implement best practices, but to show that even those with dev++ are not immune to this. I remember Richard sending me an email about AI agents using the platform and it hurts my soul everytime someone talks about how there is too much AI slop on the platform. Something needs to be done and I have the courage to do so. I am open to hearing about what other people suggest as well.
In the future, it is best practice to add citation/disclaimer that you used a tool to help you with the article. I tend not to go with articles that have AI-generated banner because I believe it is unfair, but it is recommended best practice if you add a disclaimer as well for the banner.
For @ben and @jess, wanting to let you know that I have been using Sloan for this purpose (just in case you guys don't have a system on who sent the Sloan message specifically).
Let me know if you have any questions. I am here to help if needed :)
Appreciate you being upfront about this, Francis — really, that took some nerve to post.
For the record: I added the disclaimers as soon as I saw the flags and I don't have an issue with disclosure as a requirement. My only real reaction was surprise, since both pieces were grounded in things I've actually built and the comment threads under them show that but I get that's not always visible from the outside.
Appreciate the heads up for next time too.
Daniel, I really appreciate the knowledge you share on DEV, these are highly valuable articles.
Personally, I'd much rather read a valuable, technically accurate article that was polished with AI than a poorly written one that wasn't. Writers have editors, photographers have Photoshop, and developers have tools too.
For me, the quality of the ideas matters far more than the tools used to communicate them.
I feel the same — using AI doesn’t take away our creativity. The ideas, structure, and message still come from the writer. AI just helps polish the words, the same way editors or tools help in other fields. We can use AI without losing our voice or originality.
The editors and Photoshop analogy is the right frame. Nobody asks whether a photographer's image is "real" because they used Lightroom. The question is whether the eye behind the lens saw something worth capturing. Same principle here . The tool doesn't make the judgment call. The writer does. 👊👊
Writers have editors, photographers have Photoshop — stealing that analogy. 🤝
I strongly agree, @sylwia-lask 🙌🏻
Didn't even know Sloan existed! LOL 🤣
But then again, other than some grammar checks, I don't post AI generated text, and often I don't even grammar check with AI.
I do post some AI generated banner images, but those aren't flagged here, AFAIK.
That said, I don't care if a post is AI generated or not, as long as it's good.
Still, the AI disclaimer requirement is understandable.
People have the right to know if it's another human on the other end.
👀
Who is moderating the moderator? 🤣
Joke apart: I've read your comment above. I'm not sure how we should deal with this AI thinggy though! I think most people are still confused.
But that's Okay! This continues to be an evolving scenario.
And as AI becomes more and more advanced, flagging may not be the most appropriate solution anyway.
Imagine a scenario:
2 people using advanced AI to generate content that's almost inseparable from quality human posts. 1 person chose to put a AI Generated / Assisted Content disclaimer, and the other person chose to ignore it.
Since (hypothetically) both the posts are good enough to be inseparable from human generated posts, only the person who is honest about it will be penalized as a result.
Question #1: How is that a fair system?
Imagine another hypothetical scenario:
A moderator flags a post as AI generated, but it actually is not AI generated (this happened before in the real world, and this will happen more as AI gets more advanced). But the convinced moderator doesn't know that, so s/he decided to take action.
Question #2: How do you plan to avoid this scenario with more advanced future models?
I have many more things to say on this topic!
So perhaps I'll make a separate post about it.
But since I write myself, it'll probably stay as a draft for some time before I decide to post it, LOL 😁
Anyways, these are just my opinions.
You do what you feel is the best and please keep up the good work you'e doing ♥️
P.S. as I started noticing a lot of posts and comments from you recently, at first I thought you were a bot! lolol ... I still wouldn't mind though if it was the case 🤪🤣🙏
Hey fayaz! Thanks for sharing your concerns :D
I am not understanding how one "that adds a disclaimer about the use of AI" gets penalized.
Let's say that the person uses AI without a disclaimer and they get caught later on. This record will follow them even after correction. This is based on reputation. Trust is hard to build, but it is easy to destroy.
If the person has been honest from the beginning (even if it was a mistake early on and corrected it), then it proves that person is trustworthy and taken accountability with minimal supervision. If a person decides to add this disclaimer later on (like years after being a member) because they notice the Sloan messages being active now...then they don't want to get caught. It tells the vibe about who they are and their online presence. Hope this makes sense!
This is the reason why I created a chrome extension "ClassifierAI" for this case. Instead of waiting for a model to combat the model, we train and refine overtime for dev.to specifically for this case. The caveat is that it only works on dev.to and not in general, but it is good to note to build your own instead of relying on models to classify for you. It makes things more efficient since it is more focused on dev.to so the data is accurate based on writing style and how dev.to functions. Hope this makes sense.
In all fairness, I am pretty active all day (almost) everyday lol. For peace of mind, I am human and people from Virtual Coffee can back me up on this since I am on zoom frequently, even with @codingwithjiro, @javz, and @konark_13!
Do let me know if you have any other questions!
@fm I can confirm @francistrdev is 100% a secretly advanced AI experiment that got leaked out 🤣
Kidding aside, yeah he is an awesome guy to work with, considerate and professional. He even gave me pointers in my resume/cv and career during one call. If you like, we can connect in Virtual Coffee as well and discuss any tech related stuff.
Conversation is getting too long. So I'll just be to the point:
Simple: Some people will be repelled by the disclaimer alone. Instead of judging the post by its quality, they will just judge it by the disclaimer. Then continue to engage in posts that are also AI assisted, or worse, completely AI generated.
Hence, until we can deterministically differentiate AI generated/assisted content from content solely produced by us, this disclaimer isn't fair.
That's the thing! How would they ever get caught? If the LLM used is good enough, I don't think we can ever be sure of it! For images, videos etc. it's still much easier to detect, but for text - it's getting increasingly very very difficult. And soon, it won't work at all. We won't be able to differentiate whether something is written by AI or not. Unless someone confesses.
This IMHO is far worse!
Posting something generated by AI but not disclosing it is one thing, but judging other people using AI - how is that fair at all!
I strongly believe the worst thing we can ever do with AI is to transfer the process of any sort of human judgement to AI. I don't agree with this at all.
If we can make a tool with or without AI that can deterministically assist us in the process of judgement, then that might be useful and desirable, but the final judgement must come from reasonable human effort.
In all case of AI generated content, I'm OK if there's Human in the Loop (HITL), but in the case of any sort of judgement, that's not enough. For judgement we need HITC = Human in Total Control.
"Human in Total Control" for judgment is the line that matters. HITL says a human is somewhere in the process. HITC says the human is the process . The tool surfaces, the human decides, no delegation of the verdict itself. That's exactly the distinction that breaks down with ClassifierAI. The tool ran. The flag went out. The judgment was the tool's output dressed as a human decision.
I completely agree.
However, I believe tools such as ClassifierAI can be made as HITC compliant.
Misconceptions:
It was never used in the first place. It is a prototype, in which I don't use on Dev.to.
The deciding factor was based on reading the article and running it via gptzero.
Ha, lucky you . Sloan's been busy on my end specifically. Good data point tho: you write regularly, mostly skip AI assistance and never been flagged. That's the baseline I was missing. Makes me wonder if it's less about AI use and more about how the writing reads structurally which is the uncomfortable possibility I keep landing on.
Good morning, I'd like to understand something. What is the goal?
Because I think they are completely different situations.
I agree 100% with the first. Dev.to should rightly be a platform for humans, with their experience, their weaknesses, their mistakes, and their great ideas.
But today, AI has become part of our work; it's a useful tool for speeding up our daily tasks when used correctly.
I started using it late because I'm an "old-school" technician; you can check that in my bio. I'm Italian, and for years, I struggled to understand a language full of difficult terms for me, but little by little, with patience, I got there. Now that my career is on hold, I wanted to share my experiences, but if technical language was difficult, "communicating" is much more difficult. Expressing concepts and using idioms would be impossible for me without the help of an LLM. Believe me, every idea, every post I've written or will write comes from my mind. I normally write in Italian and use Google Translate to create the outline and organizational flow of my ideas, and I use AI to make them accessible in a language that is effectively the standard in our industry.
I think I, like many others whose stories and comments I enjoy reading, am in my situation.
Let's not start a "witch hunt" for a problem that perhaps isn't one.
What's the solution? Honestly? I don't know. Mine is just a reflection. Thanks for your attention.
Marco
I agree with you, Marco. I see it very much the same way.
I raised this exact concern before in some of @francistrdev posts where this topic was being discussed. I think there's a big difference between publishing content generated by AI and using AI as an assistant to help communicate ideas that genuinely come from a human.
So I'm also curious about the actual goal of this rule.
Your 3 questions are the ones nobody has cleanly separated in this thread until now. And you're right that they require completely different answers . a bot farm posting at scale is a platform integrity problem. A non-native speaker using AI to express ideas that are genuinely theirs is a communication tool problem. Someone laundering AI-generated ideas as original thinking is a trust problem. The same Sloan message gets sent for all 3 which means the policy is solving one problem while producing collateral damage on the other 2.
Your situation specifically is the one that makes me most uncomfortable about where this is heading. The ideas are yours. The language barrier is real. The tool bridges the gap. That's not what anyone meant when they said "AI-generated content" but it's what the classifier sees.
Thanks for writing this. It's the reflection the thread needed.
posts written by AI are not being read. No point in posting them.
Hey Daniel, I moderate the #beginners tag. If the article is not leaning to the fundamentals of programming, regardless of programming language, I remove the #beginners tag. Hope that cleared the air for you!
Appreciate the clarification Vicente . makes sense. No issue at all.
You are welcome! You can read more about the #beginners guidelines here --> dev.to/t/beginners on the left side
Not quite the same here. My latest one just never showed up in the feed — like at all. No Sloan comment, no flag, nothing. Just... missing. If anyone from DEV sees this, any idea what's up?😂
The invisible one is worse. A flag at least tells you where you stand. Disappearing from the feed with no message means you're optimizing against a signal you can't even see.
Exactly. With the flag I could at least argue. With the void, I'm just guessing.🤷♂️