Why AI Writing Sounds Like AI
What language models default to, why it reads as flat, and what detectors measure.
Before you can fix AI writing, it helps to know why it sounds the way it does. This module covers the mechanics in plain English: how a model chooses each word, the two numbers detectors lean on, how the detectors themselves work, and the patterns readers notice first.
No maths is needed. By the end you'll be able to look at an AI draft and say exactly what makes it feel flat, which is most of the battle.
After this module you can
- Explain why AI text drifts toward the safest, most common phrasing
- Read a perplexity or burstiness score and know what it measures
- Judge how far to trust a detector result
- Spot the 12 most common tells in any draft
If you're short on time, read lesson 1.4 first, then come back.
- Lesson 1.1 How an AI Model Picks the Next Word A plain-English look at how language models write one word at a time, and why that habit makes AI text sound safe, smooth and generic.
- Lesson 1.2 Perplexity and Burstiness, Explained Simply Perplexity and burstiness are the two numbers most AI detectors lean on. Here is what they mean, with a chart, and how your edits change them.
- Lesson 1.3 How AI Detectors Work (and Where They Fail) Classifiers, perplexity scoring and watermarks: how AI detectors decide, how accurate they are, and why they wrongly flag human writers.
- Lesson 1.4 12 Tells That Give Away AI Writing The 12 patterns readers and detectors notice in AI writing, from stock words to lists of three, with an example and a fix for each one.