7 Comments

"It’s not so dissimilar from the humble spider web. We don’t value the beautiful and intricate designs made by a spider,"

Wait, we don't?

As far as that new chatbot goes, have you tried it out? It's incredible. As a professional writer myself, I hope you're right that we're both long dead by the time such a thing can replace us (luckily, I'm likely to go before you, so my chances are better), but honestly, I'm not sure I'm so convinced. It definitely worries me.

Maybe the difference is I'm a technical writer. Don't get me wrong, my craft is still really unique. Not a lot of people can do what I do, and I do craft a lot of content "out of thin air" so to speak. However, much of what I do is helping non-technical people understand technical concepts. I'm not so sure an AI won't be able to do that effectively quite soon.

Expand full comment

It's possible the AI will become sophisticated enough to perfectly replicate technical writing, but I imagine your job would then transition to being an AI editor. Technical writing requires a lot of precision, so I imagine most firms would not trust the AI to get it exactly correct.

I do think the chatbot is already good enough to do general content for websites, with some amount of editing and revisions done by a human. Maybe soon (possibly within even months!), the AI will get to the point where it no longer technically requires the editor, but I imagine that editor will still have a job for a number of years.

Google translate and other automated translation services have come a long, long way, but I don't think anyone will really trust it to be exactly correct for many years. But I imagine you could get the translation to be, like, 70% of the way there, so then a human editor would go through and clean it up.

But maybe this is just the optimist in me! I know any sufficiently sized company is going to try to get rid of translators and technical writers as quickly as possible in favor of a free AI service (or probably this will become a subscription service sold to companies), but I don't think we're as close to this as people might imagine.

Too, I foresee some amount of lawsuits coming over this based on what is being fed to the AI in order to teach it to put out what it puts out. Possibly to the point that it will shutdown these kinds of AI projects for a long, long time.

Expand full comment

This is a wonderful, and very humane essay. I completely agree: We humans value souls. After the novelty wears off, where is the fun or the meaning in all this machine-made junk? (And that wolf has no teeth. Evolution says thumbs-down on that!)

On a side note, your speaking voice is terrific! You could have a career as a voice artist, or, if you sing, as a basso profundo!

Expand full comment

Ha! Well, thank you. I don't think anyone would normally qualify me as a bass! But my voice does sound deeper in that recording for some reason.

Expand full comment

I probably qualify as a weird tech freak, but...I do have AI art on my walls. I printed out some landscapes I generated.

I think the issue between, say, an AI playing chess and an AI generating art is that it's always clear when an AI is playing, but AI art is already indistinguishable, in many cases, from human art. This might be the weird tech freak talking but...I don't think anyone can actually feel soul in images or text. Or perhaps better to say: they can, but it's a meaning they give to the image/text, not one that inherently exists in the art. If we curated 10 traditional paintings and 10 AI paintings, I don't think anyone could differentiate between them with any success (a common "own" in the fighting between traditional and AI artists has AI artists deliberately mislabeling AI art as traditional and traditional as AI, and baiting traditional artists into pontificating on why the "traditional"--actually AI--art is superior). I think preferring non-AI art is going to turn out to be a bit like sommeliers preferring pricier wine: the difference isn't in the product itself, but in the perception of the product.

I think there's also a "democratization" of art happening with AI art that shouldn't be overlooked. Something like this (https://youtu.be/03wXTK9HI88) isn't all there yet, but it's notable that it was achieved without needing to either spend thousands of personal hours on CGI or thousands of dollars to hire a VFX team to do it. There's at least one person running around the StableDiffusion subreddit who was an artist who suffered an accident that left him unable to draw, and is now using AI art to reclaim his abilities.

Expand full comment

On the democratization of art - I think this is mostly the other way. Art is already pretty democratic in that anyone can learn to do it. That doesn't mean they'll do it well! But they can do it. The materials are fairly inexpensive and widely available (for writing, for example, you just need a pen and paper or the phone [I do know several professional writers who write all their novels on their phones, which honestly shocks me] or computer you already own).

I also think the democratization happening here is more in the vein of how twitter was meant to democratize public discourse. I think the extent that that ever happened is pretty weak.

I don't know that people can't differentiate between AI art and professional art. I mean, if I put the image that I got from DALL-E2 next to a human's painting of a wolf, I think the difference is quite obvious.

For non-figurative art, this may be more accurate. There's already the very common joke of someone seeing an abstract piece of art, like a Pollock, and commenting on how that looks like it was done by a child. So when it comes to abstract and non-figurative art, I imagine you're more correct. Many years ago, we began seeing elephant art and the same argument about indistinguishability was used then as well: https://youtu.be/PLxO7MPUKH8

But the figurative art that the AI comes up with seems very easy to distinguish from what a human is capable of.

And I don't mean that people *feel* the humanness of the writing or painting (though I do think there is a strong element of this in what we define as artistic taste) but that most people would not want to read a novel written by an AI if they knew it was written by an AI. There's the novelty of it that would drive curiosity, but I think that's about as far as it would go: people would pick it up out of curiosity.

And while you have some AI art on your wall that you made, I think an important question would be whether you'd hang up AI art that someone you don't know created, and would you pay money for it?

Expand full comment

To the last--sure. Why wouldn't I? If I like the image, why would I not pay to have a copy of the image for my own? And for an AI novel--why wouldn't I read it, if I enjoyed the story?

It's already indistinguishable. Plenty of AI-generated stuff is obvious and bad, but then plenty of human scribblings are also bad. Peruse some of the work in the various AI art subreddits:

https://old.reddit.com/r/midjourney/comments/wsfljj/asked_the_ai_what_it_sees_as_beautiful/

https://www.reddit.com/r/midjourney/comments/z7fzn1/what_happens_when_you_mix_ancient_egypt_with/

https://old.reddit.com/r/StableDiffusion/comments/xqbgjq/a_robot_paints_a_portrait_of_greg_rutkowski/

https://old.reddit.com/r/StableDiffusion/comments/y25wve/meow/

https://www.reddit.com/gallery/zc746h

https://old.reddit.com/r/sdforall/comments/y2664x/jackolantern_icecream/

There are often tells, if you know where to look for them (pupils, hands/feet, and things like clothing that doesn't make sense). But they're often negligible, and each release the AI is getting better. Midjourney v1 to v4 represents an insane amount of growth.

I think the "democratization" part is the lack of time needed to get to the "bone", and the easing of ancillary skills. Learning how to mix paint colors, for example, used to be a critical skill for an artist: if you couldn't remember the right ratios for different colors (or couldn't afford the more expensive hues), you couldn't get far. Digital art made this skill irrelevant (among a host of others), and at the time Photoshop was taking over, there were a lot of "but artists need to learn the fundamentals/it's cheating". Nobody's saying that now, though.

Tools make things easier, and making things easier makes it more approachable and accessible to a wider range of people. The time it took me to learn how to use StableDiffusion well enough to generate portraits for a D&D campaign was infinitely less than it would have taken me to learn traditional portraiture. It was also free/used items I already owned.

Expand full comment