New AI labelling feature

I posted this in another thread but I figured this could use its own thread.

Discourse has added a new AI tagging feature for quotes, code, and details.

Quote example:

>[!AI]
>I’m sorry Dave.
>I’m afraid I can’t do that.

I’m sorry Dave.
I’m afraid I can’t do that.

Code Example:

```javascript ai
console.log(“I’m sorry Dave, I’m afraid I can’t do that.”);
```

console.log("I'm sorry Dave, I'm afraid I can't do that.");

Details example:

[details=“Summary here” ai]
I’m sorry Dave, I’m afraid I can’t do that.
[details]

Summary here

I’m sorry Dave, I’m afraid I can’t do that.

You can also get to these by clicking the options icon on the upper right (the plus sign in a dark circle) and selecting “Add AI Content” then selecting your choice from that.

I pinned it for the next 2 weeks.

Great idea

Explain like I’m five.

Does this mean that if I post a long quote from Gemini or ChatGPT or an image generated by one of them, I should use the “a.i.” tag like I would use a “trump” tag or a “celebrity-death” tag?

ETA. Re the example: preferably in the same post.

There is an AI tag that we have had for a while. You can use that to tag posts that are about AI, like you would use a Trump tag or a celebrity death that. Nothing with that has changed.

We also have the following rule, which may be found in the Terms of Service (linked to at the bottom of every SDMB page)

If you have a quote from Gemini or ChatGPT (or any other AI), you wouldn’t use the AI tag. You would somehow indicate on the post that the text you are quoting is AI generated.

For example:

if I googled “How many Rs are in strawberry” and I used Google’s AI answer, before Discourse added this feature I could post something like this:

This answer came from AI:
There are 3 "r"s in “strawberry”: strawberry.

Now, with Discourse’s new feature, you can do this instead:

There are 3 "r"s in “strawberry”: strawberry.

Note the AI tag that Discourse automatically added to the right side of the quote.

You can format it yourself by typing something like this:

>[!AI]
>There are 3 "r"s in “strawberry”: strawberry.

Or you can accomplish the same thing by clicking on the Options icon (the plus sign inside the dark circle on the top right of the composer window), select Add AI Content, then select Quote, then paste your quote over the AI Content text that it generated.

There are 3 "r"s in “strawberry”: strawberry.

Again, note that Discourse added the AI tag over on the right side of the quote.

Since the spelling of strawberry has nothing to do with AI (unless you are discussing how AI used to screw it up) you wouldn’t add the AI tag to the thread. We’re talking about a quote tag, not a thread tag.

Make sense?

Note that if you use the rich text editor in the composer window, you won’t see the actual tags, and you’ll probably need to use the Options icon to get it to work properly. Just paste your AI generated text over where it says AI Content in the quote it creates.

There are 3 "r"s in “strawberry”: strawberry.

Basically, we’ve always (well, since AI has been a thing) required Dopers to specify when they’re posting AI-generated content. This new feature just provides a simple, concise way to do that.

Or both. :folded_hands:

I think there’s some confusion with how this is worded, at least I know I was confused about it. I read the OP multiple times before I figure out what was meant by “AI Tagging”.
This is unrelated to what is normally referred to as tagging a post. All it’s doing is putting that “AI” symbol in the quote.

Aw jeez, is this gonna be another discobot-hates-Tripler fiasco . . . just now with AI?

Tripler
I don’t know if I can handle SkyNET ignoring me.

Any way to put AI-tagged posts on ignore?

No. Not yet at least.

But it is a nice easy to see visual indicator to the right side and should be easy to skip.

You can ignore a Forum, a thread Tag or a User, not a coding tag.

Steps to Mute a Tag or Category (Forum)

Click your user icon next to the menu.
Click the last Icon for Profile
Click Preferences
Click Tracking
The lower right hand box is for Muted Tags.
Enter tru and you’ll be prompted for trump. Click trump.
Click Save Changes
.
.

To Ignore a user from Preferences

Click your Profile Avatar in the upper right hand corner
Click the Profile icon (at the bottom of the icons)
Click the ⚙️ Gear icon for Preferences
Click the Users menu item
Click Add button under Ignored
Type in the username
Select a Timeframe, which is a pain on a phone.
Click Ignore

Remember, it’s not a post that’s marked as AI. It’s some portion of a post. Do you really want to hide a complete post just because that user chose to show their prep work by citing the actual AI response that provided maybe 10% of their facts and 1% of their analysis?

You do you, but that seems to want to throw an awful lot of babies out in a futile attempt to prevent ever being splashed with any bathwater.

IMHO, calling this new feature “tagging” is confusing the issue.
Maybe “AI Highlighting” or “AI labelling” or something would help.

If you really think that’s how most people are using AI (here and elsewhere), then I think you’ve been drinking the bathwater.

Thanks for this new labeling, mods - I think it will be helpful (assuming people use it.)

Two points:

  • How I described is how I use AI at SDMB.

  • The official [AI] labeling here is applied by the human author. Somebody substituting purely AI “workmanship” for their own is very unlikely to bother labeling it as such.

So proposing to block (or just not read) any post containing the new [AI] label amounts to slurping up the slop and straining out the good stuff. Not a well-considered move IMO.

But YMMV of course.

I should have called it “AI labelling”. That’s a better term. I’ll use that going forward.

I have updated the thread title accordingly.

That’s how I use it, too.*

AI is a tool, nothing more. I often find it to be a useful tool.

*And whenever I’ve used it, I’ve always labeled the information as coming from AI.

Gold star for you both! Nevertheless, my original contention stands.

But I worry we are getting off-topic for this thread, so if you’d like to discuss AI use here and elsewhere, I would be happy to join you in one of the many other AI threads that exist.