Stop Giving AI Companies an Out

While skimming Writer’s Digest, I came across this clickbait headline. “Think AI is Bad for Authors? The Worst is Yet to Come” by Mike Trigg¹. I knew I’d regret it, but I clicked through to read. I had to stop reading the article because it made me so angry that I started screaming at my phone.

One of the first complaints came from this quote:

The idea of artificial intelligence is that it mimics _human_intelligence, but at machine scale. Just like a child learning to read and write, the more the AI engine consumes, the better it understands.

But the second sentence is incorrect. The machine is never going to understand the text. But the more you feed into it, the more it will have to mimic. And thus the predictive text gets more data. Feed in one million descriptions of a rabbit, and at the end of that, if you ask it what a rabbit is, it will spit out a congealed predictive text glob describing a rabbit. But it doesn’t understand what a rabbit is. If you then showed it a photo of a rabbit, it wouldn’t respond “oh, that’s a rabbit.” But you can tell children stories about rabbits and then they will recognize rabbits they see outside. Because machines don’t understand and humans do.

But then Mike wrote the following quote, and I was annoyed enough to stop reading and write a comment in reply.

If a specific piece of copyrighted material is only a tiny fraction of the input, and that material has an indeterminable impact on the output, how would one calculate an amount due to the author or publisher of that work?

But this is incorrect on its face. Simply put, the individual who writes something and holds the copyright has the authority to determine its usage and the amount of compensation they will demand for such usage.

Although the AI companies can use misleading calculations to assert the content’s lack of value, they overlooked the initial step of seeking permission. If they’d asked, maybe some or most authors would have said “sure.” But they didn’t even ask. They sucked it all in and then shrugged, saying, “it’s only a drop in the bucket of the content we scraped.”

It’s irrelevant that we can’t prove how much the AI is using or not using. Copyright means that we control how our copy is used. And claiming fair use would only be relevant if they sucked in a few lines of every book they scanned. Because again, fair use is not determined by how much content there is as a whole, but rather by how much of that particular copyrighted work was used. And most times, the LLM companies used the entire book or corpus, not just snippets.

By posting articles like this, you appear to be in the pockets of the AI industry. And despite the title, you don’t appear to have authors’ rights or well-being in mind at all.

Don’t get me wrong, I’m sure his argument about Discovery Bias has weight and is yet another thing to be worried about. But I didn’t read through it completely, as I was still fuming from his apologist attitude towards AI companies in general.

If you enjoyed this rant and would like to get notified of future books, posts, or other mentions, join my newsletter: Dryads, Dragons, and Druids.

Image by 00luvicecream from Pixabay

• ai
• Annoyances
• rants