AI Marking for Teachers: Top Marks AI Achieves 0.95 Correlation on Edexcel English Literature - Shakespeare: 20 Mark Extract Question
Research shows Top Marks AI hitting 0.95 correlation with Edexcel for Shakespeare: 20 Mark Extract Question, November 3, 2025
AI Marking for Teachers Achieves 0.95 Correlation for Edexcel's Shakespeare: 20 Mark Extract Question
"How accurate are your AI GCSE English Literature marking tools?" It's the question that comes up time and again in our conversations with schools and teachers.
As such, we've been running a series of tests to demonstrate just how accurate the Top Marks' GCSE English Literature AI marking tools really are. The results speak for themselves!
We're examining the performance on Edexcel English Literature -- specifically, the Shakespeare: 20 Mark Extract Question.
Edexcel makes available numerous exemplar essays for their exam papers and we've put our tool to the test using 59 of those very same exam board approved standardisation materials. These exemplars showcase a broad spectrum of answer quality. These are official standardisation materials that show teachers the spectrum of answer quality.
We took 59 of these essays and ran them through our dedicated marking tool. Then we measured the correlation between the official marks the board awarded each essay, and the marks Top Marks AI assigned to those same essays.
We measured correlation using the Pearson correlation coefficient. In short:
- • A value of 1 would mean perfect correlation -- when one marker assigns a high score, the other always does too, and when one assigns a low score, the other always does too.
- • A value of 0 means no correlation whatsoever -- knowing one marker's score tells you nothing about what the other marker awarded.
- • Negative values would mean the markers systematically disagree - when one assigns high scores, the other assigns low scores.
For context, how do humans perform?
What sort of correlation do experienced human markers achieve when marking essays already marked by a lead examiner?
Cambridge Assessment conducted a rigorous study to measure precisely this. 200 GCSE English scripts - which had already been marked by a chief examiner - were sent to a team of experienced human markers. These experienced markers were not told what the chief examiner had given these scripts. Nor were they shown any annotations.
The Pearson correlation coefficient between the scores these experienced examiners gave and the chief examiner was just below 0.7. This indicated a positive correlation, though far from perfect. If you are interested, you can find the study here.
How did Top Marks AI perform?
Across all 59 essays, Top Marks recorded a correlation of 0.95 -- an incredibly strong positive correlation that far outperforms the experienced human markers in the Cambridge study. (Top Marks AI was also not privy to the "correct marks" or any annotations).
Moreover, 93.22% of the marks we gave were within 2 marks of the grade given by the chief examiner.
Another interesting metric is the Mean Absolute Error, for which our system scored 1.03. On average, the AI differed from the board by 1.03 marks, which is comfortably within 2 marks. As a percentage, that's an average of 5.2% difference.
In contrast, in that same Cambridge study, experienced examiners marking a 40-mark question showed a Mean Absolute Error of 5.64 marks, that's a difference of 14.1%. These results highlight the exceptional accuracy of Top Marks AI compared to traditional marking practices.
We don't claim that Top Marks is infallible, but when it does get things wrong, just how bad is it? Well, let's turn to the Root Mean Square Error to find out. Root Mean Square Error (RMSE) is a measure of the severity of large errors. When you square the number 1, you still get 1, and when you square 2, you still only make a small jump to 4. But square 5, and you're suddenly all the way up at 25. That's how RMSE works - it (essentially!) highlights large errors by squaring them.
Top Marks AI's Root Mean Square Error was 1.50, meaning even when larger errors occur, they remain remarkably small relative to the 20-mark scale.
You can see the full side-by-side human and AI scores below.
| Essay ID | Board Score | Top Marks AI Score | Difference |
|---|---|---|---|
| 2024 Macbeth 1 (?) (3).pdf | 3.0 | 3.0 | +0.0 |
| 2024 R&J 1 (?) (6).pdf | 6.0 | 7.0 | +1.0 |
| 2024 MOV 1 (?) (9).pdf | 9.0 | 10.0 | +1.0 |
| 2024 MAAN 1 (?) (20).pdf | 20.0 | 19.5 | -0.5 |
| 2024 Macbeth 2 (?) (8).pdf | 8.0 | 9.3 | +1.3 |
| 2024 R&J 2 (?) (12).pdf | 12.0 | 13.0 | +1.0 |
| 2024 Twelfth Night (?) (12).pdf | 12.0 | 13.0 | +1.0 |
| 2024 MOV 2 (?) (13).pdf | 13.0 | 13.0 | +0.0 |
| 2024 R&J 3 (?) (20).pdf | 20.0 | 20.0 | +0.0 |
| 2024 Macbeth 3 (?) (16).pdf | 16.0 | 16.8 | +0.8 |
| 2024 Macbeth 4 (?) (20).pdf | 20.0 | 20.0 | +0.0 |
| 2023 Twelfth Night (?) (10).pdf | 10.0 | 10.0 | +0.0 |
| 2023 R&J 1 (?) (10).pdf | 10.0 | 9.5 | -0.5 |
| 2023 MoV 1 (?) (12).pdf | 12.0 | 13.0 | +1.0 |
| 2023 Tempest 1 (?) (18).pdf | 18.0 | 16.0 | -2.0 |
| 2023 Macbeth 1 (?) (6).pdf | 6.0 | 8.0 | +2.0 |
| 2022 Twelfth Night 1 (?) (11).pdf | 11.0 | 10.0 | -1.0 |
| 2022 R&J 1 (?) (5).pdf | 5.0 | 5.5 | +0.5 |
| 2022 Macbeth 1 (?) (6).pdf | 6.0 | 5.5 | -0.5 |
| 2023 R&J 2 (?) (18).pdf | 18.0 | 19.5 | +1.5 |
| 2022 MoV 1 (?) (11).pdf | 11.0 | 13.0 | +2.0 |
| 2022 MAAN 1 (?) (20).pdf | 20.0 | 20.0 | +0.0 |
| 2023 MOV 2 (?) (20).pdf | 20.0 | 20.0 | +0.0 |
| 2021 R&J 1 (?) (20).pdf | 20.0 | 20.0 | +0.0 |
| 2023 Macbeth 2 (?) (9).pdf | 9.0 | 9.3 | +0.3 |
| 2021 Macbeth 1 (?) (15).pdf | 15.0 | 13.0 | -2.0 |
| 2020 Macbeth (?) (10).pdf | 10.0 | 10.0 | +0.0 |
| 2020 R&J 1 (?) (9).pdf | 9.0 | 13.0 | +4.0 |
| 2019 Twelfth Night (?) (12).pdf | 12.0 | 13.0 | +1.0 |
| 2022 R&J 2 (?) (14).pdf | 14.0 | 13.0 | -1.0 |
| 2019 MOV 1 (?) (18).pdf | 18.0 | 18.0 | +0.0 |
| 2019 R&J 1 (?) (10).pdf | 10.0 | 9.3 | -0.7 |
| 2022 Macbeth 2 (?) (11).pdf | 11.0 | 9.3 | -1.7 |
| 2019 Tempest (?) (15).pdf | 15.0 | 16.8 | +1.8 |
| 2023 R&J 3 (?) (20).pdf | 20.0 | 19.5 | -0.5 |
| 2022 MoV 2 (?) (16).pdf | 16.0 | 16.0 | +0.0 |
| 2019 MAAN 1 (?) (9).pdf | 9.0 | 13.0 | +4.0 |
| 2019 Macbeth 1 (?) (10).pdf | 10.0 | 10.0 | +0.0 |
| 2018 Tempest (?) (8).pdf | 8.0 | 9.3 | +1.3 |
| 2023 Macbeth 3 (?) (12).pdf | 12.0 | 13.0 | +1.0 |
| 2018 MOV 1 (?) (12).pdf | 12.0 | 13.1 | +1.1 |
| 2018 MAAN (?) (20).pdf | 20.0 | 19.5 | -0.5 |
| 2017 Tempest 1 (?) (8).pdf | 8.0 | 7.0 | -1.0 |
| 2017 Mov 1 (?) (9).pdf | 9.0 | 8.0 | -1.0 |
| 2022 R&J 3 (?) (10).pdf | 10.0 | 9.3 | -0.7 |
| 2017 Macbeth 1 (?) (8).pdf | 8.0 | 8.0 | +0.0 |
| 2017 MAAN (?) (17).pdf | 17.0 | 13.1 | -3.9 |
| 2019 R&J 2 (?) (14).pdf | 14.0 | 16.0 | +2.0 |
| 2019 Macbeth 2 (?) (20).pdf | 20.0 | 20.0 | +0.0 |
| 2023 Macbeth 4 (?) (15).pdf | 15.0 | 13.0 | -2.0 |
| 2017 R&J 2 (?) (10).pdf | 10.0 | 10.0 | +0.0 |
| 2022 Macbeth 4 (?) (18).pdf | 18.0 | 16.8 | -1.2 |
| 2018 R&J 1 (?) (9).pdf | 9.0 | 8.0 | -1.0 |
| 2022 Tempest 1 (?) (20).pdf | 20.0 | 20.0 | +0.0 |
| 2017 Twelfth Night (?) (18).pdf | 18.0 | 16.8 | -1.2 |
| 2017 Macbeth 2 (?) (18).pdf | 18.0 | 13.0 | -5.0 |
| 2017 R&J 1 (?) (15).pdf | 15.0 | 13.0 | -2.0 |
| 2022 Macbeth 3 (?) (14).pdf | 14.0 | 13.0 | -1.0 |
| 2024 Tempest 1 (?) (9).pdf | 9.0 | 9.5 | +0.5 |
Can I see a graph to help me visualise this?
Absolutely.
First, here's a scatter graph to show you what a theoretical perfect correlation of 1 would look like:
Now, let's look at the real-life graph, drawn from the data above:
On the horizontal axis, we have the grade given by the exam board. On the vertical, the grade given by Top Marks AI. The individual dots are the essays -- their position tells us both the mark given by the exam board and by Top Marks AI. You can see how closely it resembles the theoretical graph depicting perfect correlation.
Discover how Top Marks AI can revolutionise assessment in education. Contact us at info@topmarks.ai.
We use cookies to enhance your experience. By continuing to visit this site you agree to our use of cookies. Learn more in our Cookie Policy.