11/26 2022
I recently attended the tail-end of a conference about language technology, where one of the keynotes talked about generating traditional Irish folk music using ML (available here). One central point about his talk was the backlash that he faced from different parts of the internet. About how FolkRNN would --- among other things --- "corrupt the pure" or "won't lead to anything" or how it is "the proof of the beginning of the end for humanity" (sic. hyperbole added). His answers to those criticisms were good. In short, he stated that the current AI revolution of content generation simply is one further step in the line of technological evolution. I think, however, that this reasoning fails to explain the actual backlash, by which I mean it fails to explain exactly why people feel so strongly regarding the way that AI now encroaches on the creative space. It is, I think, best explained through a short quote by Dostoyevsky's 'Notes from the underground':
"The whole work of man really seems to consist in nothing but proving to himself every minute that he is a man and not a piano key."
There are two main ways to respond to this, assuming Dostoyevsky is correct here.
Now, (1) is unreasonable. Technology, historically, cannot be stopped in virtue of perceived harm (sometimes very unfortunately). That the technology will be invented is, I think, inevitable. But when it comes to things like aiding people in creative outlets I just fail to see any way to contain the benefit it could provide for some people without a total ban.
The second option concerns people changing. Now, people can change, but most won't ever do it willingly. But it is necessary, I think, in this case to attempt to do so.
What would such a change entail, you ask? A very good question, and one not easily answered. I think part of the answer lies in the realization that just because you won't ever beat someone else there is virtue in doing something for the sake itself. Writing a novel isn't about getting published, earning money and fame. It is about writing, about putting something on the page that you know and care about, in a way that best tells the particular story you want to tell. Making a movie with some friends and having a great time making it. Painting a nude portrait of your significant other, all the while having a boyish smile on your face as you perform the brush stroke, and a grin on hers. Because, all those things have one thing that AI generated content simply doesn't --- and can't ever --- have in the same manner: human stories that matter to the people in them.
The revolution that is currently happening in all creative spheres, whether that be image/video/audio or text, will change the way that we perceive ourselves in the world since it'll undermine one fundamental aspect of what we have always seen as fundamentally human: our capacity for autonomy, of proving that our beliefs are not as binary as the 0 & 1's in a computer (or the black and white keys of the piano). It is a failure, I think, to come to this conclusion when it comes to these new kinds of large machine learning models, however. The problem isn't that they for example can generate music that when played is indistinguishable from music written by a human. The issue is when one loses track of what actually matters when it comes to these creative spheres: the creative endeavor itself, which is a story in itself. And stories are the way that we connect with other people, the way that we understand other people.
Soon the flood of AI-generated content will overwhelm online marketplaces, libraries and streaming services. In that flood the era of the particular story will probably begin. Perhaps as a service providing catered AI generated content. But more probable is that services providing catered human stories will flourish, or physical spaces where creativity is analogue. Because, I think Dostoyevsky was right in the fact that for most of us, one fundamental aspect in how we define ourselves in this world is a rebellion against the very idea that we can be defined at all. For many I don't suspect this to change, no matter how large the evidence against it. And if that's true, the market will find some way to exploit that.