I’m not writing this in defense of LLMs. At this point it seems to me that their impact on the world is going to be like the Internet’s – a lot of really good stuff, and a lot of really bad stuff, all wrapped up into one impossible human development.

On one hand it’s hard to look at the toll this development has already taken – on artists whose work was first stolen and is now devalued, on older folks being manipulated by fake content, on students who are underserved by a school system that hasn’t yet adapted – and not feel a little sad.

Still, sometimes when considering this, I’m comforted by a certain line of thinking:

A friend comes to you and hands you a book. They say they just finished; it’s a great story, gripping and emotionally poignant. You should totally read it! You flip it over to skim the back cover, and it seems interesting enough. Then your friend feels obliged to tell you, in case it matters, it was written by ChatGPT.

The question being – would you still want to read the book?

Would anyone want to read the book?

If not – why not?

For the first two questions, I feel pretty safe in answering “no” for myself, and I can’t imagine I’m living in a world where many people would still be excited about reading that book.

But, for the last one – I don’t know. It seems like a hard question.

For me, I guess, the point isn’t the answer, but moreso that the question exists at all. If my assumption is right, and nearly everyone on Earth would find it more or less a waste of time… doesn’t that say something? About our shared sense of value, about what makes something feel “worth it” to us?

It implies that, no matter how convincing the LLMs get, no matter how cohesive and engaging a story they can create, we’re still looking for something else in what we read; something that can only come from someone else deciding it mattered enough to write down.

I’d guess as the years go on, and more and more of the content we run into every day is LLM-generated, this distinction could become more familiar to people. And maybe, if we get more practice at separating the “worth” from the “worthless” along this axis, we may start to notice some other things have the same flavor as LLM-generated blog posts or pictures – things like designed-by-committee AAA video games or made-for-Instagram museums.

It could lead to us wanting to be more deliberate, to know the human behind our content.

I think that could be one nice thing to come from all this.