Beyond grammar: can AI understand the nuances of academic writing?

On July 4, 2024, the free versions of ChatGPT, Gemini, and Claude were evaluated for their effectiveness in grading technical English assignments in an academic context. The study sought to determine if these AI platforms could serve as reliable grading tools by analyzing their responses to two sets...

Descripción completa

Detalles Bibliográficos
Autores: Montalbán Martínez, Nicolás, Montaner Villalba, Salvador
Tipo de recurso: artículo
Fecha de publicación:2026
País:España
Institución:Universidad de Alcalá (UAH)
Repositorio:e_Buah Biblioteca Digital Universidad de Alcalá
Idioma:español
OAI Identifier:oai:ebuah.uah.es:10017/68377
Acceso en línea:http://hdl.handle.net/10017/68377
Access Level:acceso abierto
Palabra clave:Instructional design
ESP
ICT
PBL
AI
Diseño didáctico
TIC
ABP
IA
Humanidades
Filología hispánica
Humanities
Spanish philology
Descripción
Sumario:On July 4, 2024, the free versions of ChatGPT, Gemini, and Claude were evaluated for their effectiveness in grading technical English assignments in an academic context. The study sought to determine if these AI platforms could serve as reliable grading tools by analyzing their responses to two sets of papers. The first set contained original student submissions written in technical English, while the second consisted of the same papers rewritten in grammatically correct but academically incoherent language. The AI models provided three grading suggestions for each assignment and were also asked to assign a final grade. These AI-generated grades were then compared with the grades given through traditional manual grading by educators. The analysis focused on the consistency of the AI's grading, its alignment with human assessments, and its ability to distinguish between technical precision and overall linguistic coherence. The findings contribute to ongoing discussions about the role of AI in education, particularly in the assessment of academic writing. This study highlights the potential benefits and limitations of using AI as a grading tool, emphasizing the need for further refinement before such systems can be reliably implemented in academic settings.