پويا شهيدي

This thesis investigates the effectiveness of artificial intelligence (AI), specifically GPT-4, compared to human eva‎luators in assessing argumentative writing. Using a quantitative approach, the study eva‎luates 183 graduate student essays based on key components such as claims, counterclaims, rebuttals, and supporting evidence. The results show that while GPT-4 performs efficiently in assessing primary elements like claims and data, it struggles with more complex aspects such as counterarguments and rebuttals, where human eva‎luators offer more nuanced and context-aware feedback. The research highlights significant discrepancies between AI and human scores in these areas, suggesting that while AI can enhance efficiency in large-scale assessments, it cannot yet replace human judgment in eva‎luating deeper cognitive and rhetorical structures. The study recommends hybrid eva‎luation models that combine AI’s scalability with human eva‎luators' critical thinking for more accurate writing assessments.

هوش مصنوعي , جي پي تي 4 , نوشتار استدلالي , ارزياب هاي انسان , ارزيابي نوشتاري

Artificial Intelligence (AI) , GPT-4 , argumentative writing , Human eva‎luators , Writing Assessment​

Puya Shahidi

Dr. Shahin Vaezi