Elon Musk recently changed (Nov 15, 2024) X’s (formerly Twitter) user agreement. From that point on, anything posted or displayed becomes the royalty-free but non-exclusive property of Musk to sell as he pleases, including for A.I. training purposes. This comes down to Musk not paying for your content—is he paying you now?—but also that you still own your content and may sell it to others if you wish. I noticed the summary I read did not distinguish between public and DM messages.

This policy change has naturally occasioned discussion on X, and I have said for a long time that I do not care if my content is used for A.I. training purposes. To be clear, I think people have legitimate concerns about the misuse of personal data. I may elaborate on this later, but for now, I’m only going to say that most of those “personal abuse” issues already occur; they do not need to buy what Elon is selling. In this hopefully short essay, I will only talk about A.I. training and response.

What does A.I. want with our content? A.I. is not trying to steal your work (images, text, music, what have you) to duplicate it. It is reading, viewing, or listening to your work to “understand” it (notice the scare quotes)! It wants to “know” how elements go together, what is a “proper” association or transition, what key concepts or color/line associations or musical phrases are present, and how they relate to those around them, including all the other music or images or text the A.I. “sees” in its training. That’s millions, even billions of texts, images, or bars of music. What constitutes “your work” in there is a miniscule portion of the total.

Human artists have done this with one another’s work since Homer! I do not know a writer who doesn’t read others and “draw” from what they read. The same is true for the visual or musical arts. When I write on a particular subject or craft a story, all that I have read is, in some part, melded into how I express myself—a combination of my own skills and what I have learned and remember from others. A.I. is doing exactly this when it outputs an answer. It associates around a topic using all that it has “learned.” 

Of course, A.I. has total recall and we do not. Critics would say that A.I. can’t add anything strictly novel. That may be true, but I suspect much of what we think of as novel about our own creative thought or that of others has come from some long ago encountered thing whose source—in our minds—has been forgotten. The difference between us and the A.I. is that the A.I. doesn’t forget! The human melding of past experience and A.I. melding of training data are entirely different mechanisms, but the results are similar. 

If some random human can read my tweets, essays, or novels and assimilate an idea that later finds its way into one of their works—remember, it isn’t a copy of my work—I would be honored by my inclusion. Why should it be any different if it is an A.I. that does the assimilating? 

Suppose that phrase of yours, hardly noticed today, comes to be on everyone’s lips thanks to another artist’s—or A.I.’s—output fifty years from now? Either way, no one will know that you are the original creator of the phrase or technique, but either way, your creation is made to live again! That is why I invite Elon (in fact, I dare him, but that is another matter) to use my tweets to train A.I. As for Amazon or WordPress, they can use my books, reviews, and essays! I won’t get paid, but then I’m not being paid now and I’m still writing stuff!

My books and essays are copyrighted (A.I. cannot just “read them out” without getting into abuse territory already possible without A.I.), but copyright does not prevent other authors from incorporating my thoughts—even literally, as fragments of text in their work. Why should I object if some future A.I. output does the same? That kind of fragmentary inclusion is pretty much the worst thing that could happen!

***

Why do I say above, “I dare Musk”—or Bezos, or the owners of WordPress—to let A.I. train on my writing? Because if they do dare, their A.I.’s will end up with such original phrases as “consummate conscious cunnilingists” and “pluripotent poetic porn” in their repertoire! I say let them!