This may not exactly be an AI problem, but I didn’t realize that at first.
I wrongly blamed a generally reputable chatbot.
I had just read a wonderful article on LinkedIn about subtle issues in AI bias.
It had been republished from a website that had many well-written and thought-provoking articles in the AI & Society space.
So, I asked my favorite chatbots for sites similar to that one.
None of them gave me a useful response. Except for one suggestion, I knew all the suggested sources and would count them all as reputable sites, but they did not match the topic well.
One had never heard of. It turned out to be a site with a list of AI-driven chatbots rather than articles; essentially, all of those chatbots were erotica.
I wanted to mark the choice as “stupid.”
But there isn’t a way to do that. The recommendation list included citations to a source that listed similar sites.
This particular item was wrong.
Is the whole source page unreliable?
The whole site?
The source that listed that site as reliable?
How broad a brush should I apply the “stupid” label?
I don’t know.
And even if I took the time to figure it out I couldn’t easily fix my interaction with the chatbot.
And even if I could fix it for me in a configuration file or prompt generator with ever-growing lists of sources to ignore, I wouldn’t be able to share that easily. I supposed I could make a customGPT, but even if I did that, there would be severe limitations on what it could do.
So, even if I never know who to blame, I wish I knew how to blame; how to blame in a way that it would do some good.
BTW: Apparently, direct confrontation is objectable.
ChatGPT said:
I wasn’t able to generate the image as requested due to a conflict with the content policy. If you’d like, we can explore other ways to adjust the theme while staying within guidelines. Let me know how you’d like to proceed!
You said:
adjust it to be inside the content policy
ChatGPT said:
To stay within the content policy while keeping the idea colorful and accusatory, I’ll shift the focus from direct blame toward a symbolic confrontation.
