I haven’t really talked about AI on here, have I? It’s not because I don’t have opinions on the subject, believe you me. It just…hasn’t come up, I guess.

But now, it has.

The Atlantic posted a search engine based around Library Genesis, the data Meta (Facebook’s parent company) is using to train its AI. As an author, you can go to the page and type in your name and see if they’ve used any of your books to help train said AI.

And they have:

So, let’s talk about AI for a second, shall we?

I am…not a fan of AI in practice or concept. I recognize that there are a bunch of technocratic twerps who are trying desperately to shove their multi-billion dollar boondoggles down our throats with these things, but I just do not want to use them. They are erratic, inaccurate, and soulless.

Worse yet, AI is built on theft. AI “art” is just scraped images from other, real artists that the algorithm smushes together like a five year old with a Barbie and a G.I. Joe shouting, “Kiss! Kiss!” “Oh, but it levels the playing field, allowing those who have no artistic talent to create art!” some folks may be shouting. Counterpoint: you know what else allows you to create art? Taking the time to develop the talent for it. Talent in artistic endeavors is not a thing that I think really matters all that much. What matters is mastering the discipline and practice necessary to become good at doing something. There’s that old saying in the gym, “No pain, no gain,” and I think it applies to any sort of effort or product. If you want something to come out good – whether it’s a book, a song, a drawing, a report you made for work, a meal you cooked – you have to put in the effort to make it good. That means some of your early efforts are gonna be absolute dogshit (please no one dig up my drawings from high school. They are terrible and no one wants to see them. Or my old poetry, for that matter).

AI does not “create” anything. It recombines what already exists, often in the most boring way possible. Creating takes conscience, which AI (as it exists now) does not have. Will it in the future? I don’t know. But I know that “training” AI on stolen works is not the way to go about doing things. So, shame on Meta. I’ll be waiting for my check.