2025-03-03 8:40 AM
I couldn't help noticing that many posts lately have keywords and/or whole phrases highlighted in bold typeface. This is an entirely new phenomenon, I don't remember seeing this previously, or at least not in such conspicuous proportion.
Can please somebody enlighten me, why?
Maybe it's not a bad thing, I can't decide yet.
Thanks,
JW
2025-03-03 8:57 AM
This, for example?
I'm not sure it's entirely new, but certainly the prevalence has greatly increased.
No idea why - shame people can't similarly learn to use the code formatting ...
2025-03-03 10:00 AM
>> shame people can't similarly learn to use the code formatting ...
Or communicate useful detail for problem diagnosis..
It's not just here, so perhaps via AI chat or translation. I find it visually distracting.
2025-03-03 10:07 AM
@Tesla DeLorean wrote:>> shame people can't similarly learn to use the code formatting ...
Or communicate useful detail for problem diagnosis..
Indeed.
I've also noticed an increase in the number of posts with a great many words, often with many headings & sub-headings, yet conveying very little useful information.
It does, indeed, smell of AI.
2025-03-03 10:50 AM
I use bold for submitter text extracted that I am responding to.
2025-03-03 12:11 PM
I don't think that's what @waclawek.jan is talking about here?
More like this example:
On the forums broken 'quote' feature:
https://community.st.com/t5/feedback-forum/klunky-quoting/m-p/631751
https://community.st.com/t5/feedback-forum/how-to-properly-use-the-quote-tool/td-p/755735
2025-03-03 12:20 PM - edited 2025-03-03 12:52 PM
>>I use bold for submitter text extracted that I am responding to
And that's fine, I think the complaint here is random high-light/emphasis and walls of text.
I would encourage participants here to use the forms that work for them, even using native language where they can express themselves most fully, as the translation options have got so much better. Photos of hand drawn diagrams, also very workable means of conveying details and flow.
End-users have historically presented briefly without a lot of formatting, some of the recent posts are looking more mechanically generated.
I'm not sure we're going to change behaviour, and use of AI for first level support ideas and check-lists, but rather just confirming Jan's and Andrew's observation
2025-03-04 12:43 AM
Hello @waclawek.jan, all
Thanks to brining this to our attention. We will definitely observe this new behavior.
Whenever you feel the content is predominantly alerted by AI, feel free to report it to our review.
For reference: Why predominantly AI-generated posts and replies on the ST Community are not allowed
2025-03-04 2:35 AM
Hi @Lina_DABASINSKAITE ,
I don't think it's AI-generated. As @Andrew Neil and @Tesla DeLorean have discussed above, it's more like AI *translated* and/or - and more importantly - doctored.
In itself it's not harmful, it just unnecessarily dilutes the post and may attract false attention to unimportant items, although the posters probably think otherwise.
JW
2025-03-04 5:53 AM
I've noticed this as well. Given the fairly high number of nonnative English users on the forum, I assume it's some sort of rewriting of input from ChatGPT. Many of these follow the same bullet point structure. Honestly, seems better than the "X doesn't work, why?" types of posts.
I've also noticed a much larger proportion of register-only code coming from users (example) where the code is unnaturally well-commented using /* comments */. I suspect the same thing. ChatGPT spits something out, doesn't work, user posts it here.