In a time where educational institutions worldwide seek to balance objectivity and efficiency, while integrating new technologies in teaching and learning, the potential role of artificial intelligence (AI) has become both promising and controversial. This paper addresses such pressing concerns by examining how students perceive AI-driven grading, identifying both its strengths and limitations through the lens of a structured online debate.

This paper presents the results of a thematic analysis performed on qualitative data collected through a voluntary reflective exercise integrated into an introductory, self-paced AI course at an engineering university. The exercise invited students to participate in a structured online debate, focusing specifically on the role of AI in assessing student work. Participants were prompted to articulate their views on whether AI should be responsible for grading and evaluating students’ assignments, emphasizing its potential to eliminate human bias while acknowledging possible negative impacts on creativity and individuality.

Through the application of thematic analysis, a qualitative research method designed to identify, analyze, and report patterns within data, several key themes were identified. Students generally expressed an appreciation for the objectivity and efficiency AI could bring to certain assessment tasks, such as multiple-choice questions, mathematics, and plagiarism detection. They acknowledged the potential of AI to relieve teachers from monotonous grading tasks, thus enabling educators to devote more time to higher-order pedagogical activities.

However, a significant number of participants voiced strong concerns regarding AI’s limitations, particularly in evaluating creative and reflective assignments. Participants emphasized the irreplaceable role of human empathy, nuanced understanding, and interpretive capability in assessing tasks requiring subjective judgement, personal reflection, or emotional insight. There was broad consensus that AI currently cannot adequately capture the depth and complexity inherent in tasks involving creativity, critical thinking, or emotional nuance.

A prominent perspective highlighted by students was the proposal of a balanced hybrid model, where AI initially assesses straightforward tasks or identifies potential issues such as plagiarism, followed by human oversight to ensure accuracy and fairness in more complex evaluations. The analysis indicates a preference among students for AI as an assistive, rather than autonomous, assessment tool.

The findings suggest a cautious but optimistic approach toward integrating AI in education, advocating for a balanced strategy where AI augments but does not replace traditional pedagogical methods. This paper contributes valuable insights for educators, researchers, and policymakers interested in leveraging AI to enhance educational practices responsibly, preserving the richness of human-led learning experiences. At the same time, it sets the tone for future-oriented discussions that involve AI as a key player in educational processes, either administrative or purely educational, since it includes student voices that may be less advantaged when matters of policy or design are at hand.

Keywords: Artificial intelligence, assessment, e-learning, higher education, automation.