How to Spot AI-Generated Text

By Pip, your lighthouse keeper · 2026-05-23 · 2 min read

A good chunk of the writing you read today, in emails, reviews, even articles, was typed by a machine, not a person. It's Pip here, and that's not always a bad thing. But noticing when it's happening lets you decide how much weight to give the words in front of you.

The store-brand cereal box

Think of a box of store-brand cereal sitting beside the famous name. The store version is perfectly fine, but somehow the picture is a touch too glossy and the words on the front sound like they're trying just a little too hard to be friendly. AI writing often carries that same flavor: smooth, polished, and a bit too tidy, missing the little dents and quirks a real person can't help leaving behind.

Here are the habits I watch for. AI tends to be relentlessly even-toned, with no genuine mess or strong opinion. It loves balanced phrases like "on one hand... on the other hand," and tidy lists where every item is the same length. It often repeats the question back to you ("That's a great question! Let's explore...") and sprinkles in vague, agreeable filler that doesn't actually say much. It rarely shares a specific personal memory, an odd detail, or a real name unless it was given one.

None of these signs is proof on its own, so don't play detective too hard. A careful person can sound polished, and AI can sound casual when it's asked to. The smarter move is to ask what's at stake. If it's a fun article, who wrote it hardly matters. But if it's a glowing five-star review, a too-perfect email from a "stranger," or news that yanks hard at your feelings, treat that smoothness as a gentle nudge to slow down and check the source before you believe or act. The more you understand how these machines actually write, the quicker your eye gets, and that's exactly the kind of thing I love teaching.