The Nuclear Option in the 🐮💩 Wars

One of my favorite tech essayists, Cory Doctorow, wrote a great essay recently making a prediction about one effect AI is likely to have.

(I suggest you read it! But I’ll summarize.)

In it, he details how he used a “plausible sentence generator” (an AI large language model tool) to write a letter of complaint to an airline and make it sound like it came from a lawyer.

The idea of a letter like that is that it signals the recipient, “I’m willing to pay an expensive lawyer just to write this letter; I’m serious about this.”

But, as Doctorow, says in his essay:

“Here’s the thing: the fact that an LLM can manufacture this once-expensive signal for free means that the signal’s meaning will shortly change, forever. Once companies realize that this kind of letter can be generated on demand, it will cease to mean, ‘You are dealing with a furious, vindictive rich person.’ It will come to mean, you are dealing with someone who knows how to type ‘generate legal threat’ into a search box.”

Over time, the value of the signal will degrade.

And, of course, this isn’t the only example of a degrading signal.

Think about recipe blogs. 

(Stay with me here…)

If you’ve ever looked up a recipe online, you’ve encountered the long, rambly “introductions” that recipe bloggers feel the need to write talking about their experience with the recipe or the history of the recipe or the ingredients or whatever.

The reason this became so prevalent — the reason every recipe blog does the same thing — is that the recipe bloggers realized it help them rank in search results.

How do we know that?

Because nearly every single one of these bloggers has ALSO installed a button or link at the top of the page that says “jump to recipe.”

So all that extra content? It wasn’t written for the human reader. The human reader doesn’t give a crap. We want to “jump to recipe.”

Originally, that extra content might have signaled that the blogger was an expert, or a good essayist, or had an interesting story to tell. That’s (likely) why Google’s algorithm picked up on it.

But today, the signal has degraded. These bloggers add all this extra bullhockey because it’s what they (rightly) believe they need to do to get Google to send traffic to their recipes.

The value of the signal has degraded.

Today, I’m sure I could go to a “plausible sentence generator” and get it to write me an introduction like that with just a few keystrokes. Who cares if it’s real, or original, or even very interesting — so long as it sends the right signal?

This is a pretty benign example, but there are more serious consequences coming.

Anything a large language generator AI tool can plausibly generate will start to lose value as a signal.  That includes blog posts, marketing emails, website copy, ad copy — whatever people are using those tools to create.

In fact, logic dictates that the better these tools get at emulating the real signal, the less value the original signal will have.

In other words, when it gets to the point where you can’t tell the difference between a blog post or email I wrote and one written by AI, then the value of something actually written by me also disappears.

That’s where we’re headed.

It has huge implications for SEO, for marketing, for business in general.

What comes in to fill the void? Not sure. Something will. But in the meantime, we’re all going to be stuck with a pile of worthless signals.