About two years ago, the Internet had a heart attack when they heard the former head of the self-driving car division at Google and Uber was creating a Church that would worship an AI Godhead. The story ran the press circuit for about a week and then everyone forgot.
Circling back around to the topic about two weeks ago, I was on an Inevitable/Human Digital Hangout to hear what other future thinkers were thinking about the topic of AI religion. One of these future thinkers, Russell, joined the hangout and we discussed the concept of using AI to rewrite the Bible.
Upon doing some research, we quickly found that we weren’t the only creators inspired by this prompt. So rather than reinvent the wheel, I thought I’d echo some of the coolest findings.
Already a future thinker?
Then become a friend.
Rewriting the Bible
John Lynch, showcased on his website IV.AI, that he trained an AI to write new verses of the King James Bible. The results could’ve been easily transplanted into the Bible I’ve read many times (almost) without noticing:
13|12| Abram dwelled in the land, beside the first seven fat kine:
29|13| And it shall come to pass, when he overthrew those cities, and that which grew upon the cattle should be alone; I will bless thee, and will give unto thee, and make mention of me unto this day.
Similarly, Tobias Hermann recreated scriptures of the Bible, however, he turned it into an interesting game called Divine or Benign. It’s a game in which you are given a scripture and must determine whether it’s a real passage or one generated by AI. It’s quite fun.
For all the developers that want to test the waters themselves, Nilesh Kumar developed an entire “How To” manual for other programmers to create their own version of the Bible using a GAN (Generative Adversarial Network). Sleeba Paul also detailed his entire process of thinking that went into recreating one book in the Bible, which he would call the “Gospel of LSTMs” – LSTM stands for Long Short-Term Memory, a type of algorithm variant of RNN (Recurrent Neural Network).
However, in my opinion the coolest conclusion to come out of these experiments was Elias Mahanna on the Fast Forward podcast who theorized the idea of a religious matchmaking service:
If an algorithm can tell me what kind of music I’m likely to want to listen to based on my previous listening habits on Pandora, why can’t it also tell me what sort of faith system would be best for me?
Let’s say if somebody were to say, “This year I’m going to read these 10 religious texts, and I’m going to read them in a format that I devise where I can essentially mark them up, using semantic markup. And I’m going to give a thumbs up to paragraphs or verses that I find compelling and interesting, and thumbs down to things that I don’t like…
[An] algorithm could then go through that entire body and cobble together a very useful sort of “Amazon suggests” these 5000 words as the core of your own religion.
Undoubtedly, there was a lot of shock value to come from this whole line of a God-like AI.
Personally, I think the idea of a unified religious text that brings together the passages of all religions is fascinating. The only thing I worry about, though, is which charismatic leader would get their hands on this powerful text, how they would use it, and what hidden intentions they had.