Just because we can doesn’t mean we should

In our ds106 noontime coffee chat yesterday we talked a little about the potential impact of AI on art. This made me think of the Folk RNN project from a few years ago. They trained a neural network on a database of fiddle tunes and had it generate new tunes and titles. I remember finding one, “The Drunken Pint.” The title caught my attention as it indicates that the app recognized a relationship between words but not the meaning. I tried to play the tune, but it didn’t sound right with my limited skill, so I asked an actual musician to try it. Something about it still didn’t sound right. In all likelihood it’s just me, but most of the computer generated tunes sounded off. I think it’s the quick beats, the triplets and sixteenth notes, and their placement that does it. They sometimes feel like a Max Headroom style hiccup or stutter.

I stumbled upon a video on the Folk RNN backstory:

This piece below the video caught my eye. It’s the new magic number:I liked the part about ethics, and how it only occurred to them afterwards. I suspect that this is not unusual. People decide to do things because they can, without thinking about whether or not they should. I remember hearing someone say about Zuckerberg that many people could have built what he did, but they all had the good sense not to. Writers and storytellers have a role here in contemplating and illustrating the possibilities. What could go wrong? So much of AI in fiction tends to be dystopian, which suggests that a lot could go wrong. And most of those tales were written before  people started thinking about the ecological/environmental impacts.

And in a bit of serendipity, I also saw this yesterday:

Maria Popova is a good one to follow on Mastodon, if you’re interested in ongoing enlightenment. The article, Susan Sontag on Storytelling, What It Means to Be a Good Human Being, and Her Advice to Writers, is actually from a few years ago, but was highlighted because it was Sontag’s birthday. The article referenced, At the Same Time: The Novelist and Moral Reasoning is in the Files section of Canvas because I thought it might be a good reading for the course, as it questions some of the potentials of digital storytelling and the web. Maybe those questions could be applied to AI as well.