Tag Archives: technology

This is yesterday’s issue of my newsletter Time’s Corner.

The other day my brother showed me some videos created by Google’s Veo 3 tool. Watching them, I felt like the Spartan king seeing a catapult for the first time: “Here is the death of valor.” What’s the point of struggling to make a beautiful film or portrait or poem if any rube with a computer can do the same thing?

The thing is, I actually like making movies and taking pictures and writing poems. I don’t just like having them. I like making them. At the end of the day, I’m not too worried about Veo 3 personally. So, I’m going to ask a more neutral question: What are these tools for?

Like all tools, AI is decided to make work easier, usually by making it more efficient. They reduce friction. Is there any benefit to reducing friction in creative work?

Alan Jacobs wrote about this a few years ago. The friction between the artist’s ideal and the limits of his tools is a key factor in creating art. Artists need their materials to push back. (This is what always bothered me about that video of Glen Keane animating in virtual reality. “You can do anything you want.” Yeah, but what I want is to not be able to do anything I want.) On the one hand, artists don’t want to make their work more efficient. But, on the other hand, artists can and do adopt new tools. Someone had to invent the lead pencil, after all.

I suppose the answer to my question depends on what you’re going for. In the filmmaking industry, digital video has almost entirely replaced actual film because it’s gobs easier to record, manage, edit, and alter. If you want to make a great-looking movie as quickly and cheaply as possible, digital is the way to go.

But not everyone wants to do that. I recently met a filmmaker here in Birmingham who loves using DVR cameras (remember those?). He makes music videos for bands who want a certain, unconventional look, and DVR gives him that look. (He also mentioned that people behave way differently in front of a camcorder than they do in front of a phone. Pull out a phone and everyone turns away. Turn on a camcorder and they all start mugging and crowding to get into the frame.)

I do most of my writing on the computer because it’s fast and I like the semi-formal look of type on a white background. When I need to slow down and think through a scene or paragraph, I put out pen and paper. (Not pencil. Too easy to erase.) But most of the time, I’m after speed, and the computer gives me that. Could you use an LLM in the same way?

Cory Doctorow has written that his main problem with AI art is the lack of communication behind it. Art communicates via the million micro-decisions made by the artist in creating it. Each stroke of paint (or the keyboard) tells something about the artist’s thoughts. You can create AI art without making very many decisions; hence, according to Doctorow, its uncanniness.

But what if there is intention behind it? If an artist uses Dall-E, or whatever, to make hundreds of versions of his idea, then chooses from among those hundreds, makes adjustments, generates a hundred more, etc. If a poet has ChatGPT spit out a poem about the rain, then changes a word, then another word, then another, chops the whole apart and puts it back together? Does that make it just like any other tool, one step beyond using a word processor?

Let’s say I’m writing a scene and I’m stuck. “What would Jane say in response to Michael’s comment?” I wonder. Being the old-fashioned writer I am, I would do one of the following:

1. Lean back and stare at the wall for a while
2. Grab a piece of paper and a pen and start writing one bad line after another, in the hopes of jostling loose a good idea
3. Go for a walk

It’s very easy to imagine a younger writer developing a different process. “What would Jane say to Michael?” she wonders. She opens ChatGPT, types in Michael’s line, and asks it to generate fifty different things Jane could say in response. She skims the output, selects the best option, copies, pastes, and continues on her way.

There’s much more to say, but this newsletter is already long. Let me add one thing: The problem with using AI tools to make art is that they trick you into thinking that they are creating something new. They aren’t. What an AI spits out may seem new, but it’s an amalgamation of vast quantities of words or images created by human beings. Now, my grad school professors would probably say, “What’s the difference?” All of us our simply parroting stuff we’ve heard or read. We’re just super-advanced LLMs made out of meat. But that’s simply not true. Human beings can come up with new things. (They can make new humans, after all, each with an individual soul.) Machines can’t.