The Value of Human-Created Art Keeps Going Up

For years I’ve been saying here and elsewhere that the emergence of AI in the creative fields is going to increase the value of human-created art, particularly writing, music, and even the (physical) visual arts.

At the same time, it’s decreasing the value of what we could say was already slop—LinkedIn posts, a great deal of the self-help market (there will be a big differentiator here), books made purely just for marketing a brand, content farms, social media spammers, etc.

Yesterday Substack joined the rising tide of platforms looking to solve the influx of AI content through labeling what is and what isn’t machine made. The music streaming services have been going this way. Bandcamp were the first to outright ban AI content on the platform. Tidal, Qobuz, Apple have been employing different versions of labeling the content. Now the writing platforms are hopping on board.

The platforms are now clearly betting on the human horse.

This is an obvious move for Substack, who wants to differentiate itself in the market as a place to find quality writing and thoughtful takes—or at least look like they’re trying to do that. Here’s their post below if you want to read it all, but I’ll provide a quick summary further below. I’ll then answer some common objections I saw all yesterday and this morning which are worth thinking about.

Basically, Substack have integrated the service called Pangram into the platform. Users will be able to run a quick check on anything 100 words or more, and it’ll give the article a rating on how much it seems to be written by AI.

In addition, writers on the platform can make use of a “How I Make This” feature, which really gives an opportunity for transparency and nuance.

Writers can also run a scan before publishing, and dispute if they see their writing being flagged inappropriately. Apparently writers can even turn the tool off, but I haven’t found where this is.

Going forward, users may eventually be able to filter out AI content, if they want.

That’s the summary. But why have they implemented this? Their reason(s):

When content made by no one takes over parts of the internet that are supposed to be human, it pollutes the commons and makes it hard to discover and hear human voices. When readers have to wonder if what they’re reading is real, it undermines trust in authorship and threatens the livelihood of writers—including those who use AI tools thoughtfully to produce work they believe in. Platforms that reward fakeness will create a race to the bottom.

It’s obvious, and I don’t know why it’s taken so long to get here, but most people don’t actually want to primarily engage with content, but with people. That’s what’s being admitted here. Robot-made content does not carry enough meaning for most people.

What makes an artwork (a painting, music, writing) valuable is not just the basic aesthetics but also its meaning. The meaning is not just derived from the art itself but from the way it was made, who made it, and how it is presented. In 1964, the communication theorist, Marshall McLuhan coined the term “the medium is the message”. The content of a message is one thing, but the character of it another. The platform you deliver the message on, and the way you construct it, carries a great sense of meaning as well.

In other words, it’s not just what you say, but also how you say it.

That’s why a lot of punk music, even if technically bad, can still be popular: because it has another layer of meaning. A band like Joy Division played into this contradiction exceptionally well, working hard to make music that sounds lazy, in a sense, and to craft a look that was much the same. Yet it carried multiple layers of meaning.

Writing with AI might produce what looks like perfect writing (hint: there’s no such thing, actually) but that doesn’t matter. Using it sends a message to the reader. For many, that message is something like, “I didn’t bother to spend time on this, because you don’t really matter to me, I only want to pretend you do so I can get something from you.”

You’re basically asking people to invest more time in your writing than you did. Most people feel cheated in that exchange. They don’t think of writing (or music, or paintings) in that way. Using AI sends a message, regardless of how good the prose might even be.

Objectively, jazz is more technically musical than punk music, but that doesn’t make it better, only different. People have their preferences because of what each means to them. You can be technically good at art, but there’s a lot more than just that. There’s the human story behind the art that gives it meaning. I unpacked this in more detail in this post:

People want you, not a robot, when they want to truly engage with your art.

It all comes down to what it is you’re trying to do: create content or art.

There is a difference. In 1996, Bill Gates wrote that “content is king”. That’s been the mantra for the internet. It shaped the way we engage with creativity online and in many ways industrialized creativity, particularly writing, by incorporating scale into the equation. The printed press already started that, but the internet really ramped it up—and it went into full Hulk mode with social media.

In some ways, AI might undo this by now making the differentiation between content and art much clearer. People feel it’s inauthentic if you call yourself a thought leader and writer but you didn’t put much thought or effort in. It makes you a “poser”, whether you like it or not. If you’re claiming to sell high-end ramen but are actually using two-minute noodles, no one is actually impressed.

If you want to play in the field of cheaper writing (AI) then that’s the field you should be rated against. That’s only fair and allows people to know what they’re really getting when they invest in you. Yes, there are people who use AI and then really work on it. But it’s still not the same thing as not using AI at all. It only makes sense that we know who is doing what, especially on a platform like Substack.

I have friends who use AI to write. I have no problem with it most of the time because they are not writers. They use writing in their business because they need to communicate. Training manuals, emails… a ton of things where quality prose is not really necessary in their line of work. That’s fine as far as I’m concerned, although it also has its limits. A CEO should consider how authentic they are really being if they use AI to write all their communication. But that’s a whole different story.

We’ve come through a long phase with the internet where writers have had to really struggle against thought leaders and big podcasters with massive teams and budgets that help them get content out at scale on social media and other spaces, playing the quantity game where many writers could not. Publishers jumped on board this too, and everything has been about platform size for a very long time now. AI is the end point of that game. And it may even be the undoing of it. Stacks of ‘follows’ means very little when three quarters of it may be bots. People are also getting tired of it all, and for good reason, and therein lies an opportunity for writers with an authentic voice to be valued for their craft, their voice, and their story, again.

Perhaps this is giving us all the opportunity to get off the treadmill.

I started experimenting with AI quite early on—I think it was around 2022 or 2023. I knew it would only get better, as all tech does, but I also had a feeling that people would reject the Matrix sooner or later. There are aspects of human nature that are timeless, and it doesn’t matter how great your tech is, human nature will not change. The future remains human.

I’ve seen some common objections to categorizing AI content that are popping up again in light of Substack’s announcement. Other platforms will follow. I’ve given this a lot of thought for a long time so here are some answers that I think add to the conversation. (This isn’t a defense of Substack in particular, but more a defense of the general approach of labeling AI content.)

1. Isn’t it ironic that we’re using AI to detect AI?

Yes, I guess it takes one to know one!

2. What about false positives?

This came up a lot during Bandcamp’s announcement too.

I think Substack’s “How I Make This” feature is interesting, as it allows for a creator to be open about their AI use and address the nuances.

What this could create is a culture of transparency—which is what we need—where those who might be trying to take advantage of the preference for human created writing will eventually be exposed as frauds.

If your content consistently shows up on the AI meter but you claim you don’t use it, then there will be questions.

If you’re trying to hide how you use the tool, then you should ask yourself some questions.

In my opinion as an editor—if your writing consistently shows up on AI detectors as AI written and you don’t use AI to write, then use it as a tool to address your style. Chances are your writing may be boring! (Harsh? Yes, editors can be harsh. I’m grateful for editors in helping shape my writing!)

3. I know the AI has been trained on my content, and now my writing will be flagged when it’s not AI. That’s not fair!

The AI is designed to venture toward averages. That’s partly why it can be boring, even if you try and direct it toward a certain style. This means I actually don’t think people need to worry. I know the AI has ingested my writing too, and I am honestly not concerned I’ll get flagged.

4. Isn’t AI just a neutral tool? We don’t ask other people what tools they used—the end product is all that matters.

While it’s true that every tool can be used for good and bad things, it’s also true that no tool is strictly neutral—it was designed with a specific purpose in mind. The AI industry has made it clear what they think the purpose of AI is, and they are developing it in accordance with their vision. They have told us themselves that they see an opportunity to monopolize markets; create dependency; cheapen production, to name a few. We get to decide if we want to be on board with all that.

Not to mention that they have essentially hoovered up everyone else’s content for free and then are trying to sell it all back to us, repackaged. The medium is the message, and by using AI extensively in your own production, you inadvertently put yourself into this ethics conversation and signal something to the user about what you think of it. Each should follow their own conscience, of course, but the point is to be aware of what it is you might be saying even if you’re not saying it.

What about open source / open weight models? These are developed and given away for free, running locally on computers. The ethics around this I think is a little easier to work through. However, it’s nuanced.

At the end of the day, you have to decide to pick your own battles. The point is to just know what to expect going in.

5. It’s dehumanizing to be put into a box

Sure, but yet I can bet you’re happy to niche your own art and try and market to a particular type of person. This is about market differentiation, not putting people in a box.

I know that art is intensely personal and it’s only natural to feel personally affronted. I have felt personally affronted by AI itself. However, I think the best way forward is simply to lean into the idea of niche. If you make AI-assisted art, that’s part of your niche now. It’s no use complaining that people who don’t like it don’t like it. As you grow in your art, you realize you can’t please everyone, and nor should you want to. If you were pleasing everyone, you’d please no one, because your message would pretty much be shallow and boring.

Incidentally, this is what makes AI writing pretty boring—it’s designed to try and please everyone, and so it says pretty much nothing of substance when you really scrutinize it. This is also because AI produces signifiers without any experience of what is being signified, but that’s for another whole discussion.

These are the top objections I keep seeing pop up. I think it’s good news and the way culture is shaping up around this is, in my opinion, going in a good direction.