Beware Pre-AI Value Extrapolation

The other day, a post of mine featured a Simpsons-style cartoon of me hiking, courtesy of ChatGPT's new image generation feature.

On the other hand, the first search result on "What would it cost for a graphic designer to turn my photo into a cartoon" linked to someone's Etsy shop where they offer this service for roughly $30.

So, did I just a) save myself $30 and b) cheat someone out of their $30?

No, and no.

If the only way for me to get a Simpsons-style photo was to pay someone $30, it wouldn't have been worth it. I would only bother making this cartoon if it was free.

However, others care enough about turning their photos into cartoons to pay $30 or that seller on Etsy would have no business.

Here's the critical part: Armed with ChatGPT, I could now set up a shop on Etsy and sell cartoonification for $30 a pop at almost no cost. Pure profit if you subtract the little time it would cost me to load their picture into ChatGPT and enter a simple prompt. So, a million-dollar business right here? I could even undercut the seller a bit, charging only $20 per cartoon.

I hope this example makes it painfully obvious why that won't work. As of last week, everyone who wants their photo cartoonified can now do so at a ridiculously low price point. That seller on Etsy will have to figure out something new and quick. But everyone who thinks setting up some automation in Zapier that goes from customer order to ChatGPT and back would net them $30 (or anywhere close to that) is deluding themselves.

If you have a goose that lays golden eggs, you're rich. If everyone has a goose that lays golden eggs, nobody cares about gold any longer.

In slightly more subtle ways, this will play out for many products and services. So if you build a quick-and-dirty AI automation that does X, and X used to cost $Y, chances are you can't sell X for $Y.