Some people have so much faith in artificial intelligence that they believe it represents a code-based god. Others have argued it’s another over-hyped tech that doesn’t live up to its sky-high expectations.
If you’re in the mood for some Friday blasphemy and want to decide for yourself, play the “Divine or Benign” quiz, developed by developer Tobias Hermann in 2016. The quiz feeds you ten images of text on a Dead Sea Scrolls-type aesthetic backdrop. Your job is to determine if a line is actually from the Bible, or if it’s just nonsense generated by a neural network.
According to the website, the "Benign" text is generated using a recurrent neural network. This is a kind of computing system that digitally attempts to mimic the brain and uses “loops” that make it useful for analyzing language. Neural networks learn to produce new outputs—like text, or an image—after being “trained” on a large amount of input data. In this case, the Bible.
After getting the first question wrong (I incorrectly attributed the text to the neural network), I realized a convenient winning strategy: the coherent sentences are probably from the Bible. Text that has a verb with an out-of-place “-eth” is probably the neural network trying to speak like an ancient writer.
This made me wonder: is it disrespectful to “train” a neural network on the Bible, or use them to produce fake religious texts? There isn’t exactly a Bible verse that tells you if a neural network is blasphemous or not.
Still, as someone who went to a Catholic school for three years and had to read and annotate the Bible for class, I thought that this tool has an interesting side effect: it makes you read the Bible carefully and critically, to understand it. But if you’re not religious, like me, a game like this won’t suddenly change that.
In any event, it’s just amusing to see a neural network literally try to play god.