Have you ever wondered if a machine can correctly assess your proficiency in English writing? Well, artificial intelligence (AI) tools for essay grading claim to do it. However, a recent study suggests that these tools might be discriminating against non-native English speakers, leading to inflated scores for native speakers while unfairly penalizing non-natives.
The study analyzed three AI essay-grading tools: the e-rater by Educational Testing Service, the IntelliMetric by Pearson, and the Writing Mentor by Measurement Incorporated. Researchers from the American Institutes for Research found that these tools tended to prioritize the syntax and vocabulary of a text over its purpose, coherence, and quality of argumentation. Consequently, students who write more complex but grammatically flawed sentences in English might receive lower grades than those who convey simpler but clearer ideas, even if the latter contains spelling or punctuation errors.
- For instance, the sentence "Throughout my life I had dealt with several issues so now I am well prepared to face anything" gets an excellent quality score from the e-rater, but it contains a comma splice and improper capitalization.
- On the other hand, the sentence "It's hard to imagine our world without fossil fuels" receives a mediocre score because it lacks sophisticated vocabulary and sentence structures, even though it is a thoughtful and well-organized argument.
The AI tools' bias towards native-like English might disadvantage non-native speakers who have a better understanding of the subject matter but struggle with grammar and idiom usage. Moreover, these tools might create a hegemony of the English language in academic writing and prevent multilingual students from expressing themselves fully in their native tongues.
Conclusion
In conclusion, AI essay-grading tools can provide quick and standardized feedback to students, but they should not be the only measure of English proficiency, especially for non-native speakers. Teachers and human graders can bring a more nuanced and contextualized assessment that considers the content, purpose, and audience of an essay, as well as the writer's language background and cultural perspectives. Therefore, we need to tackle the linguistic and cultural biases of AI tools to make them more inclusive and fair for all learners.
Social
Share on Twitter Share on LinkedIn