Revisit discourse's AI detection feature

[ulgens]

Code of Conduct

• I agree to follow Django's Code of Conduct

Idea Description

Revisit discourse's AI detection feature to understand how successful it is and ensure it's not hiding non-AI content

Which existing platforms would this idea affect?

• Forum
• Discord
• Github
• Elsewhere (please tell us in the description)

Rationale

The issue happens on https://forum.djangoproject.com/

Problem

I was doing a search on forum for "pyupgrade" keyword and realized that discourse automatically hides the content it think is AI. For this specific search

https://forum.djangoproject.com/search?q=pyupgrade

The forum thinks there are only 4 non-AI results, but when I checked the hidden posts, none of them appear to be AI-generated. Weirdly enough, none of these results has the search keyword in it. It looks like multiple bugs happening at once.

Would you like to be involved in this idea?

• I would like to be involved in progressing this idea.
• I am only suggesting this idea, but I do not want to be involved.

Other Contributors

Forum moderators/admins or any other person with access to the forum's settings

Other Details

No response

[github-actions]

Thank you ulgens for sharing your idea! We have a lot of them so please be patient. You can see the current queue here.

Community instructions

For commenters, please use the emoji reactions on the issue to express support, and/or concern easily. Please use the comments to ask questions or contribute knowledge about the idea. It is unhelpful to post comments of "I'd love this" or "What's the state of this?"

Reaction Guide

• 👍 This is something I would use
• 👎 This is something that would cause problems for me or the Django community
• 😕 I’m indifferent to this
• 🎉 This is an easy win
• 🚀 I would actively volunteer to make this idea happen

[ulgens]

I think I misunderstood what that filter means.

https://meta.discourse.org/t/discourse-ai-ai-search/282420

It's not filtering out the content it thinks is AI, filters out the results itself produced with the help of an AI (that is failing too). I'm not sure why we have that feature enabled, but i think it can be revisited considering the amount of false positives it generates.

That being said, this issue itself is invalid now. We can delete it if that's an option.