And there is a crucial point… Sources. With out reasonable sources for provided information, it is opinion at best.
I strongly believe that, moving forward, any posts sourced by an AI should be noted/labelled as such. I would also suggest a limit on how long AI generated info can be. This would at least allow readers to easily recognize AI content in posts and weigh accordingly.
If something like this were instituted as policy here, can the TB post editor be made to show a little message strip reminding members about flagging any known AI content in their posts?
There are sites that have decided to completely ban any AI re-posting, and for good reason. Others have had challenging times, especially when the site is a little too cozy with ChatGPT…
I understand that AI models may be trained on Stack Overflow content, but based on how often I find errors of fact if AI-generated answers, I am now discounting the value of recent Stack Overflow answers. A great way to destroy a brand!
It could—there’s some text that appears in the post editor with instructions/reminders. But let’s see if this becomes an issue. If I have to plink a few people every now and then, it’s not a big deal.
A pity that this wasn’t nipped in the bud when AI first became readily available on the internet. There could have been mandatory embedded characters in the text they generate to indicate AI creation.
Maybe ASCII 26 (substitute) would be suitable?
I’m not perturbed about AI-edited comments. A lot of posts would be radically improved with a little editing, and people clearly aren’t doing it on their own.
I’d also be in favor of an AI feature that would suggest better titles after a post was written.
