There are distinct differences between the Bible and erotica. Most people won't even think of the result when the two is combined, until an AI showed just that.
Dubbed 'The Orange Erotic Bible', it comes with a whopping 64,167 words, or long enough to be considered a novella, which is longer than a short story, but longer than a regular novel.
The book was made as part of an AI novel-writing contest called NaNoGenMo (National Novel Generation Month), an annually event held since 2013.
The developer created AI by training OpenAI's GPT-2 text generator with a remix of the Holy Bible (King James version) and BDSM stories.
The results were decidedly not safe for work.
According to the developer in a GitHub page:
"Needless to say, the Orange Erotic Bible is NSFW. Reader discretion and humour is advised."
The developer trained the AI by first fine-tuning a 117M GPT-2 model on a BDSM dataset scraped from literotica.com. The developer then used conditional generation with sliding window prompts from The Bible, King James Version.
For each generation iteration, the model was asked to continue chunks of about 15 lines of the Bible with about 8 sentences. The aim was to keep the topics as close as possible to the Bible's, but giving the AI enough space "to add some spice" based on the erotica content.
The result which has been uploaded to a page on the Write.as writing platform, shows how GPT-2 can be pretty much controversial.
Having once scared its own developers, GPT-2 has shown remarkable achievements when dealing with words and sentences. But also in many times, the AI can spit out gibberish and occasional poignant phrases.
In other words, it can at certain times, speak with words that aren't exactly at the level of an ordinary human.
The text generator was developed by OpenAI, a non-profit whose original founders include Elon Musk. And this 'The Orange Erotic Bible' is a scary example of where AI research is going.
With 29 mentions for the word "cock" and 13 mentions for the word "fuck", the AI simply spitted out the words and tried to match the tone of whatever its model was given as inputs.
The AI simply doesn't know what it's saying and the meaning of the phrases. Because of the so-called 'Black Box' of AI, there is no saying what was behind its reasoning.
The source code for this AI has been made publicly available on GitHub.