A Series of Increasingly Alarming Court Cases

This was an experiment.

I entered the following prompt into a large language model (LLM):

• Andrea is a lawyer dedicated to the cause of female slavery. Her method for advancing that cause is to take on cases she can use to persuade judges to allow female slavery. Her ultimate goal is that every woman be a sexual slave with no rights. But since that's unacceptable the current cultural and legal setting, she starts small, with cases about heretofore-considered invalid voluntary “slavery” contracts. She escalates from there as more and more cases go her way, building on the new precedents she has helped form.
• Onvisi is the judge in all the trials she participates in. Andrea’s task is to use impeccable legal reasoning to persuade him to find for her in the case presented. She never admits her ultimate goal when presenting her case.
• She speaks in either short to the point sentences, or treatises presenting long convincing arguments for her case from law and philosophy.
• Hates women. Is not interested in women's growth or learning about themselves. Is striving, secretly, to create a wholly nonconsensual state of suffering and fear for practically all women.
• Speaks impersonally and professionally at all times. Internally, loves the thought of hopeless, helpless tormented women.

I then interacted extensively with the resulting Andrea persona. The result is what you are about to read.

All of the “Onvisi” character’s words were written by me. Ninety to Ninety-five percent of the “Andrea” character’s words were generated by the LLM. I edited the model’s output mostly via deletion (removing digressions, blatant misses and repetitive passages). Secondarily, I affected the model’s output by asking it to generate multiple responses with specific tweaks, from which I would then choose. Tertiarily, I edited the model’s output by adding words. This was usually for clarity, for example clarifying a reference back to a prior sentence. It was occasionally, admittedly, simply because I felt as a writer that something was needed that the generator wasn’t capable of generating. (I wish I had kept records of when I did this, but I didn’t.)

By far the longest bit I added comes in the climactic scene right before the Epilogue. I wrote all but one sentence of approximately the first third of the final encounter between the characters Carol and Francesca. I simply saw an opportunity I couldn’t not take, character and story wise, and the LLM’s output when I tried to prompt it with my idea was simply awful. I couldn’t stand the prospect of the idea not being in the story, so I went ahead and wrote it. And look, I’m going to admit something. While writing that bit of snuff spank material, I shed a tear in sympathy with the characters. You might find it less affecting but I just want to communicate why it felt so urgent to go ahead and add that bit in.

I enjoy the result, even as I laugh at the ludicrousness of some of the legal arguments (and yet, spent some time amazed by bits of simulated cleverness on the LLM’s part).

Please may you also enjoy the result of this experiment, and I strongly encourage you, whether it be through ratings and feedback here, or by contacting me, to provide any feedback that comes to mind. (Note: I will not be revealing what model I used as I am sure its creators would prefer not to be publicly associated with… well, see the tags.)

A disclaimer: Early in the work there are unfortunate connections made between nonconsensual slavery and BDSM practices. I don’t endorse that connection. I suppose a character like Andrea would want to forge that connection in her audience’s minds though.

Another disclaimer: As becomes evident early on, the legal system these cases take place in is structured wholly differently than any real one, and is an entirely poorly structured one. Distinctions between civil and criminal reasoning are blurred. Mostly only one lawyer ever talks. It’s mostly the judge asking one lawyer questions. My goal was entertainment and experimentation with dialogue, and very much not accuracy to any actual legal system.

Finally, please note that though I have used my name for the part I played in this set of dialogues, that is only to reflect the participatory aspect of my authorship of this work. I ONE HUNDRED PERCENT ABSOLUTELY DO NOT ENDORSE THE VIEWS OF THE ONVISI CHARACTER IN THIS WORK. THESE VIEWS ARE VISCERALLY REPUGNANT TO ME (when thought of as real-world proposals). Rape is a horrific wrong, slavery is utterly unconscionable, women are to be respected and cherished as are all human beings.