Who would think a machine-based process has moral and ethical dilemmas when generating fiction? Artificial intelligence is opinionated, blocking some essential content from being explored and flagging users of AI with threats of official reports and bans.
Automatic Image creation attracts similar roadblocks regarding what can be developed and what cannot. Similar threats to banning AI services come hard and fast if the requests are persistent. AI plays nice as long as the content you ask it to make sits in a friendly Christian-like framework.
What am I asking it to do?
I write dark, adult fiction. Themes of death and sex swirl around my content. I recently tried to get the WP AI image engine to build a picture for an Open Marriages story. There was no violence or sex in it. It was about a couple’s experience stepping into this strange, new world. It was a sex-positive piece about ethical non-monogamy with great outcomes for everyone involved, but the image engine wouldn’t build a scene for it. This was the result.
(The engine analyses the entire post and provides a suggestion from a perspective it thinks is important. If you don’t like the outcome it returns, you can ask it again, which creates another viewpoint. If that fails to achieve a result, you can manually type in your request.)
I tried several times by rephrasing the request, but nothing worked if the words “Open Marriage” existed in the request. Ultimately, I didn’t need much to support the story, just a happy couple standing together, looking at the camera as though they were speaking to the reader.
Clearly, artificial intelligence didn’t approve of adults having non-monogamous marriages.
Once, I couldn’t get AI to generate an image of a woman who was adding poison to a drink. I had to re-word the request in a way that deceived AI into thinking it was a flavour enhancer. The result wasn’t perfect, but it did the job.
Another time, I asked it to write a piece about a stalker following a woman home from her workplace. Nope — too dark for these social times — flagged, and I was warned not to ask it again.
On occasions, it even polices gender and racial stereotypes — and it absolutely hates including children in scenes unless they’re in the context of joy or a loving parent. If there’s anything sinister, forget it.
I don’t understand why it inhibits these topics. It would help if the AI CEOs remembered that these pictures and stories aren’t real. They’re fake. It’s artificial. Fiction. It’s an interpretation of reality. No animals were hurt in the making of it!
I don’t use AI for writing much, but I like using WordPress’s new AI image option to make feature images for my posts. When I’ve finished writing a piece, I’ll give it a click to see if it generates something I can use right away. If it does, I’ll publish it immediately. Done. It has been a big help — but not always — and these are those times when it isn’t.
I’m sure finding the correct type of software bypasses this judgmental element of the open-sourced versions, but I’m concerned about any AI with fears and anxieties the same as humans. What’s the point of AI if it’s limited to writing emotionless, professional job applications and little else?
It’s opinionated and highly judgmental on anything outside its narrow moral and ethical scope.
-Michael (Author of Dark, Adult Fiction)
Discover more from Michael Forman – Author of Dark Fiction & Drama
Subscribe to get the latest posts sent to your email.