I Compared Toolyzo and Scribbr Using the Same 5 Text Samples. Here's What Happened.

Every AI detector claims to be accurate.

But what happens when you give two detectors the exact same text?

That's what I wanted to find out.

Instead of relying on marketing claims, I created five different test cases:

Each sample was pasted into Toolyzo and Scribbr under the same conditions.

The results were interesting.

Both detectors correctly identified completely AI-generated and very casual human-written content.

The biggest difference appeared when I tested lightly edited AI text.

One detector continued to identify its AI origin, while the other classified it as fully human.

Another surprise came from the technical AI sample, where the opposite happened.

One tool produced an exact match, while the other leaned toward AI but with a lower confidence score.

The goal of this comparison wasn't to declare a universal winner.

Five samples cannot prove which detector is "the best."

Instead, they demonstrate something far more important:

Different AI detectors behave differently depending on the writing style.

That's why relying on a single detector can sometimes produce misleading conclusions.

If you're interested in seeing every sample, screenshots, methodology, and the complete results table, I've documented the entire experiment here:

I'm planning to compare more AI detectors using the exact same methodology so the results stay consistent across every test.

4 views