NPR.
Ned Hayes is a novelist.
He's written eight books.
And a few years ago, he was reading to a book group in Portland, Oregon.
Ned was narrating his novel he had labored over, The Eagle Tree.
And someone at the reading said well, you know, I could probably push a button and create a novel this good.
I bet Ned wanted to push a button and have a trap door open.
But also to Ned, this didn't quite make sense.
Saying a chatbot could write a book misses the point of art.
The fact that a writer spent several years ideating on a book and thinking about it and plotting and writing graphs and throwing them away.
That's the love of writing and reading.
And when you're reading something, you're reading that journey.
You're reading for the backstory.
So, in a world of more and more AI art, Ned wanted to know how someone could prove to the world that they created their artworks themselves.
This is The Indicator from Planet Money.
I'm Darian Woods.
And I'm Waylon Wong.
Today on the show, artists versus AI slop.
We tell the story of how Ned Hayes took that comment from that reading group and devised a way to help human artists stand out in a world of automated art.
To learn about tech standards, we can go back to the mid-1990s.
Back then, the online community had a problem.
A web developer would code a website, but they'd have to do it differently for each type of browser.
One line to make a table, say for Netscape Navigator, and another line for Internet Explorer.
What the internet needed was standards.
So a group called the World Wide Web Consortium got together and decreed the coding rulebooks that people should use.
One of those early rulebooks was called XML, and one of the XML developers was Ned Hayes, the fiction writer.
The XML standard is used for all sorts of content and financial transactions today.
Yeah, Ned is a bit of a Renaissance man.
Along with his writing, he's also worked at Adobe, at Microsoft, and at Intel.
And Ned saw up close how important it was to have a standard, a common language that everybody uses.
Fast forward to 2023.
Ned is mulling over that comment in his reading group and he has a spark of inspiration about standards and AI.
It comes in a flash where I have the concept of the entire thing, from A to Z, from soup to nuts, and I knew there was a standard that needed to be there.
The standard would be like a seal of humanness, like a fair trade label or an organic label.
100% human.
Yeah, and like with XML, the checking process would be standardized.
Every human artist would go through the same vetting and every art consumer could trust the certification.
The first thing Ned did was build a team.
Ned named the company Human Intelligence.
Now, there are dozens of AI detectors out there.
You know, they'll scan for those long dashes that chatbots love, M dashes.
They'll look for sentences that say, it's not X, it's Y. But they're not always reliable.
So Ned knew his human certification system needed to go a level deeper.
It needed to look at the artist's drafts.
From his work with Microsoft, Ned knew that Microsoft Word and Google Docs are watching you.
That's right.
They're surveilling every keystroke you're making.
So Ned's company built a system that looks into your original documents.
It reads the metadata in your manuscript individually to see if there were any times when you, you know, suspiciously pasted 800 words of workmanlike yet beige prose into your story.
Human intelligence has a similar principle for other art, like game designers and filmmakers.
Show us the drafts.
And because Ned wanted this to be a standard, he was in this kind of catch-22.
To be widely accepted, he needed lots of users.
And to have lots of users, he needed to be widely accepted.
So Ned and his team went on a publicity spree.
We started going to book fairs.
We started going to book readings.
We started calling up all of our friends who were artists and writers and saying what do you think of this?
Do you think this is important?
And early on, everybody said, oh, that AI stuff is crap.
It's not going to work.
So we don't care.
But Ned could smell the AI slop nearing on the horizon.
The artists maybe just didn't have refined enough nostrils to smell it coming.
Take a big whiff of AI slop.
It's kind of like the smell of rancid aluminum foil.
I thought you were going to say printer toner.
I do like that smell.
Yeah, maybe it's that too.
Anyway, through his writing circles, Ned managed to sign up someone fairly high profile.
My name is Corin Tucker and I'm a musician for the band Slater Kinney.
Yeah, Slater Kinney, the feminist funk band most famous for this song, Modern Girl.
When Ned asked if she wanted to verify herself as a human artist, Corin said it was a no-brainer.
Artists care and they want people to know the difference between using a computer to write your music or write your book.
It's a huge difference.
Koren sees Ned's work as part of a wider dialogue that needs to happen between artists and AI companies.
And if we don't have agency in this conversation, I think we'll be losing a lot.
Corin thinks there's a big risk that a lot of art will be controlled by a small number of people in Silicon Valley.
So Corin went through Ned's whole certification process.
First, she showed proof of her identity.
Then, under penalty of perjury, she had to attest that she was a real human.
So then, Corinne got a little logo.
She could use a capital H with a circle around it that verifies she's human.
And now any new work would go through another process, through human intelligence, to confirm that it too is human made.
Like for a new record, she could send photos of her scrawled lyrics, early demos, and the final mix.
And then, if Corin passes, she could print that capital H logo on that new album's cover.
The business model is that the artist pays human intelligence to run the verification and keep the database of human verified works.
The hope is that the H symbol means that the song or the novel could fetch a higher price.
And a little unscientific polling shows that a lot of people would pay more.
Thank you so much.
We at the show have been around the continent promoting the new Planet Money book.
Well, we just wanted to say hello and welcome to the Planet Money Live show.
I'm Waylon Wong.
And I'm Darren Woods.
Darian and I were in Toronto.
We asked the audience what they think about all of this.
Put your hands up if you would pay less for something if you learned that it was generated by AI.
Okay.
Pretty large, overwhelming, almost consensus.
Yeah, there were a lot of hands up in the room and the artists are taking note.
Ned says that all those artists who didn't think AI art would be much of a thing back in 2023 have now started to call him asking how they can sign up.
When human intelligence launched in late 2025, Ned had 5000 artists signed up, and he expects that number to grow significantly.
Now, we don't know if human intelligence's H symbol will be the standard.
Like, the Authors Guild has its own human-authored certification.
In fact, we don't even know if there will be a standard or just lots of companies verifying human-made work.
Either way, artists are the ones bearing this cost, both in money and time to get the work verified.
The current fee for human intelligence is $20.
Yeah, so we wanted to go through the process ourselves and see how burdensome it might be for artists and see if we would pass Ned's test.
Ned got human intelligence's product lead at the time, Colette Alexander, to give us a Zoom call.
We are talking today because I think we're going to verify that this Planet Muddy episode is made by humans.
We promise we're not AI.
I don't know if pleading is effective or not.
It didn't work.
You're like, I promise I am not a robot.
If you yell loud enough, it couldn't possibly be AI.
Well, Colette had a different system, let's just say.
She told us to send over all of our raw interviews and drafts as proof that we had labored and rewritten and typed painstakingly to create the show.
I don't know why I'm feeling nervous that I might not be human.
My output's been suspiciously high and vanilla recently.
Oh, that beige, yeah, workmanlike prose.
No, it hasn't.
It's very unique.
So we emailed through our files all the raw interview audio and Google Docs.
Over the next week, human intelligence combed through it, measured the speed of our script revisions and then about a week later, we received received an email saying we failed.
No, I'm just kidding.
We passed.
We passed.
We did it.
The humans are fighting back.
This episode was produced by Julia Ritchie and Cher Vincent with engineering by Maggie Luther.
It was fact checked by Sarah Juarez.
Kate Kincannon edits the show and The Indicator is a production of NPR.
