Writing / AI
AI

AI didn't create mediocre content. It made it faster.

When people call something "AI slop," ask them: compared to what? The average B2B blog post was already generic. The quality bar was on the floor before the models showed up.

On this page

“AI slop” is the insult of the year. It’s also a dodge. It lets people blame the tool for a problem that predates it by a decade.

Compared to what?

The average B2B blog post in 2019 was a 1,200-word restatement of the first page of Google, written to a keyword, edited by nobody, and read by almost no one. It was slop. It just took a contractor two days to produce instead of two minutes.

AI didn’t lower the quality bar. It revealed how low the bar already was.

What changed isn’t the ceiling of quality. It’s the floor of effort. Generic content used to cost time, so scarcity hid the mediocrity. Now it costs nothing, so the mediocrity is everywhere at once and impossible to ignore.

Where the leverage actually moved

If anyone can generate competent-sounding paragraphs, competent-sounding paragraphs are worthless. The value moves to the things a model can’t fake:

  • A point of view that comes from having done the work.
  • Specific numbers from a real system you operated.
  • The judgment to kill the eighty percent that shouldn’t ship.

The machines made output free. That makes taste the entire game.

Related reading: score yourself with the matching audit · start with an audit · read the manifesto · I deleted 140,000 visitors a month on purpose · Customer Retention Metrics: What to Track and What to Ignore · Internal Communications for GTM Teams: How to Stop Saying the Same Thing Five Different Ways

NT
Nathan Thompson
Practitioner, not a guru. I built the growth engine at Copy.ai from scratch, then left to build Systems-Led Growth: the system that runs a company's go-to-market with one operator instead of a department. I document what I build.
Start with an audit →
Barely Shipping

I build the whole thing in public.

The podcast and newsletter where I show the frameworks, the real numbers, and the parts that don't work yet. No hustle-culture, no fluff.