Comparisons

Academi.cx vs QuillBot Accuracy Test

We compared each detector’s overall AI scores and sentence-level highlights with Turnitin’s results across five essays.

Brayden · 2 min read

QuillBot helps writers paraphrase and refine their work, and its AI detector claims to distinguish fully AI-generated writing from human text polished with AI assistance. For students using writing tools, that distinction matters. We examine how QuillBot’s judgments compare with Turnitin’s and whether QuillBot or Academi.cx provides a closer match.

Watch the full comparison on YouTube

How the test worked

Five essays were sourced from IvyPanda, a free essay database with no affiliation to Academi.cx. Selected paragraphs in each essay were rewritten using ChatGPT to introduce AI-generated content. The same version of each document was then submitted to Academi.cx, QuillBot, and Turnitin. We compared how closely Academi.cx and QuillBot matched Turnitin’s overall AI scores and sentence-level highlights.

The results from each detector

AI scores for five documents. Document 1: Academi.cx 38%, QuillBot 0%, Turnitin 36%; predicted range 31 to 42%. Document 2: 71%, 0%, and 78%; range 62 to 76%. Document 3: 50%, 0%, and 27%; range 43 to 54%. Document 4: Academi.cx 25%, QuillBot 11%, Turnitin 21%; range 17 to 29%. Document 5: 36%, 0%, and 24%; range 30 to 40%. Blue circles show Academi.cx, green squares show QuillBot, navy diamonds show Turnitin, and capped blue lines show predicted Turnitin ranges. Scores are labeled directly without leader dashes.
  • Document 1: Academi.cx detected 38%, predicting 31 to 42%. Turnitin's actual score was 36%, inside the range. QuillBot returned 0% AI.
  • Document 2: Academi.cx detected 71%, predicting 62 to 76%. Turnitin's actual score was 78%, above the range. QuillBot returned 0% AI.
  • Document 3: Academi.cx detected 50%, predicting 43 to 54%. Turnitin's actual score was 27%, below the range. QuillBot returned 0% AI.
  • Document 4: Academi.cx detected 25%, predicting 17 to 29%. Turnitin's actual score was 21%, inside the range. QuillBot detected 11% AI, 10 points below Turnitin.
  • Document 5: Academi.cx detected 36%, predicting 30 to 40%. Turnitin's actual score was 24%, below the range. QuillBot returned 0% AI.

An honest look at the misses

Three of the five documents landed outside Academi.cx's predicted range this round, which is more than usual and worth mentioning. Document three in particular ran wide, predicting 43 to 54% on a document Turnitin scored at 27%. Shorter documents are more sensitive to a handful of missed or added sentences swinging the final percentage, which the tighter documents in this set didn't run into as much. Even on the documents where the percentage missed, the sentence level highlights stayed close to what Turnitin actually flagged, which is the part that tells a student what to go back and change.

Average deviation from Turnitin across all five documents: Academi.cx 9.6 percentage points; QuillBot 35 percentage points. Values are mean absolute differences. Bars share a zero baseline and are proportional.

QuillBot barely detected anything

QuillBot returned 0% AI on four of the five documents, including essays with substantial AI-generated content clearly flagged by both Academi.cx and Turnitin. Only the fourth document produced any detection at all, and even then it caught just a fraction of what Turnitin flagged. For a tool marketed as an AI detector, missing AI content entirely on four out of five essays is hard to overlook, regardless of how the percentage predictions on the other side of the comparison turned out.

The final scorecard

Academi.cx won all five rounds. Its percentage predictions weren't perfect this round, missing on three of the five documents, but its sentence level detection remained consistent throughout. QuillBot, by comparison, failed to register any meaningful AI detection on four of the five documents tested.

Related reading