AI-content on F1-TECHNICAL

[lio007]

As the last couple of days have shown: AI-garbage is flooding the internet and it's hard to read the threads.

Do you think a forum should provide a functionality to delete or hide the original and it's associated posts? I would welcome it.

[JonoNic]

Hide and deleting all that content? You will just give the team more work to police AI generated content. A better idea would be better to create a dedicated thread to dump all AI related until it can be confirmed as authentic and move it back to the correct thread.

[sucof]

Yes, please remove them all. There are many people here who either quote AI bots texts, or just use AI to create their own comment.
I think, you could try some automatic scanning for AI comments, and also banning users for x time when caught, of course after informing everyone about this new rule.

[Vappy]

We shouldn't be using poor AI systems to generate information when it's often incorrect and not accurate. Rather than allowing it in its current unreliable state, it should be disallowed until they're proven more reliable.

[FrukostScones]

Kill it with fire!

[Apexseal157]

Yh, Like Jono said maybe the best solution is to dump all the suspect images in an AI thread until the images are confirmed, none of us want to be deceived by charltans looking to drum up hype.

[lio007]

Hide and deleting all that content? You will just give the team more work to police AI generated content. A better idea would be better to create a dedicated thread to dump all AI related until it can be confirmed as authentic and move it back to the correct thread.

Hiding in the sense when I know the posted image / video is AI-created, that I as a user can click a button and all associated posts get hidden. So it's up to the user if the AI-content is shown or not.

[hollus]

I wish we could stop calling it AI and call it LLM? Or Transformer Thrash for images?
The A is too real, the I too hyped.
But what to do about a user quoting a long, apparently convincing text from X that tests to “100% AI generated”? Ban that user?
He didn’t know that his engine would run at 16.3:1 when hot…
For now I say… downvote away?

[hollus]

Maybe a call to common sense:
a) please don AI post
b) If you want to quote text that looks like a powerpoint presentation… it is AI, please don’t post it?
One can hope…

[bluechris]

Maybe a call to common sense:
a) please don AI post
b) If you want to quote text that looks like a powerpoint presentation… it is AI, please don’t post it?
One can hope…

+1 to that. I think slowly the forum will find it's way with this in due time and users will learn to not post these things. It will need time for sure but it's not only here this problem, its everywhere.

[Richard C]

My suggestion is... AI content is allowed, but only if flagged/labelled as AI. Not that AI content is desired, but it can be posted. What is a good example of why it should be allowed? Someone can post an image that people might be talking about and saying... "This is not accurate, it is AI". For those that post content that is AI generated or highly suspected to be AI generated, but don't label as such, then moderators can discuss with them. It may be by accident. As the content gets better and better, it will be harder and harder to determine what is real and what is not. Everyone may eventually be caught out. If it is pervasive, (accidental or not) then the moderators have tools at their disposal to deal with the offending poster.

Richard

[catent]

Generative AI is essentially just an LLM, and thus is only as effective as the prompts it is given and the context it is provided.

In that same vein, "AI detection" strikes me as nonsense, given it cannot be guaranteed to accurately parse what is and is not AI. Generative AI text isn't some independent, unique entity; AI has learned how to communicate based on loads of learnings via human speech/writings and thus is effectively an artificial rendering of human speech/writing patterns.

What ultimately matters is is the quality and the accuracy of the information contained within. Just like when people went from learning new things from books to the internet, or when people went from writing words with a quill feather to a keyboard, the technology changes, but the validity of the information being communicated is still the end-all-be-all.

[WardenOfTheNorth]

I wish we could stop calling it AI and call it LLM? Or Transformer Thrash for images?
The A is too real, the I too hyped.
But what to do about a user quoting a long, apparently convincing text from X that tests to “100% AI generated”? Ban that user?
He didn’t know that his engine would run at 16.3:1 when hot…
For now I say… downvote away?

Also, I know of numerous people who have conditions like dyslexia or who are not native English speakers who have taken to using "AI" to help them write comments etc.

[hollus]

An LLM correcting a text’s style is not going to make stuff up. Similarly, an LLM summarizing a long text should not need to make stuff up.

But an LLM expanding a prompt or answering a question can and will make stuff up. Differently at different times. This is easy to test, they made my soccer team win the Spanish cup, twice, in different ways, down to the minute of the 3 goals in the final, scored by players with ages differing by 50 years.

So your suggested dyslexic scenario is a “legit” and useful usage case. As is a grammar-style check-improvement. The latter is as lazy as it is disrespectful of other’s time, IMO.

A very good point on how an automatic AI checker might fail, other than it being unreliable.

[catent]

I wish we could stop calling it AI and call it LLM? Or Transformer Thrash for images?
The A is too real, the I too hyped.
But what to do about a user quoting a long, apparently convincing text from X that tests to “100% AI generated”? Ban that user?
He didn’t know that his engine would run at 16.3:1 when hot…
For now I say… downvote away?

Also, I know of numerous people who have conditions like dyslexia or who are not native English speakers who have taken to using "AI" to help them write comments etc.

I can't speak as to whether or not AI translation (in its current form) is effective and picks up on the cultural and linguistic nuances of language, but I could certainly see it potentially being helpful in this regard.