The one thing that can save traditional media?
-
I wonder how the samples were collected and how they decided which ones were written by AI.
Based on how you do this, I guess you can get any split you want, from 1% AI to 99% AI.
@Klaus said in The one thing that can save traditional media?:
I wonder how the samples were collected and how they decided which ones were written by AI.
Based on how you do this, I guess you can get any split you want, from 1% AI to 99% AI.
No idea. There are tools professors use to determine if writing is AI that supposedly work well.
If you don't take it, it can only good happen.
-
Yeah tools out there are pretty good. I've tested them, and they provide a % AI score and I've found it's oddly accurate even if you try and manually adjust the content to make it seem less robotic. I think this is a free site where you can paste content in and get an analysis.
-
The problem with that is they are sampling things that have already been created. Once AI is most all of what's out there to sample you get into this content death spiral.
"You cannot subsidize irresponsibility and expect people to become more responsible." — Thomas Sowell
-
Is there any way of telling how many of the surveys of AI usage are performed by AI?
-
@Klaus said in The one thing that can save traditional media?:
I wonder how the samples were collected and how they decided which ones were written by AI.
Based on how you do this, I guess you can get any split you want, from 1% AI to 99% AI.
No idea. There are tools professors use to determine if writing is AI that supposedly work well.
@jon-nyc said in The one thing that can save traditional media?:
@Klaus said in The one thing that can save traditional media?:
I wonder how the samples were collected and how they decided which ones were written by AI.
Based on how you do this, I guess you can get any split you want, from 1% AI to 99% AI.
No idea. There are tools professors use to determine if writing is AI that supposedly work well.
I don't think they work well and I think they can never work well in principle.
-
Maybe not but they can work well in practice in a moment in time. For example when 90% of the students are using the same model (ChatGPT 4, say) and are all pretty prompt-naive.
I would imagine your average college English teacher was quite good at it in 2023 and 2024.
-
The problem with that is they are sampling things that have already been created. Once AI is most all of what's out there to sample you get into this content death spiral.
@Mik said in The one thing that can save traditional media?:
The problem with that is they are sampling things that have already been created. Once AI is most all of what's out there to sample you get into this content death spiral.
Funny, I was just thinking this morning how content could get into its own self-referencing vortex.
-
@Mik said in The one thing that can save traditional media?:
The problem with that is they are sampling things that have already been created. Once AI is most all of what's out there to sample you get into this content death spiral.
Funny, I was just thinking this morning how content could get into its own self-referencing vortex.
@89th said in The one thing that can save traditional media?:
@Mik said in The one thing that can save traditional media?:
The problem with that is they are sampling things that have already been created. Once AI is most all of what's out there to sample you get into this content death spiral.
Funny, I was just thinking this morning how content could get into its own self-referencing vortex.
It seems like a solvable problem. Not too worried about that one.
