Perceptions and Ethical-Pedagogical Challenges of University Professors in the use of Generative Artificial Intelligence in Academic Evaluation

Abstract

The objective of this study was: To analyze the perceptions and ethical-pedagogical challenges of university teachers in the use of generative artificial intelligence in academic evaluation. Therefore, the following theoretical nodes are addressed. Methodologically, it ventures into the quantitative approach, assuming a descriptive level of transactional cut, a population of 4,200 university actors was assumed, distributed in 1,500 students and 2,700 teachers. Questionnaires and standardized tests were used as techniques. The validity of the content was located at > 0.85 and the reliability was estimated by Cronbach's alpha coefficient, determining 0.912 for the general scale. As results, a marked asymmetry of technological adoption was obtained. 78.4% of students (n = 682, grouping options 4 and 5) continuously use generative AI to prepare academic papers, with an overall mean of X̄ = 4.12 (SD = 0.81), which incorporates the trend in the range of "Agree". In dissonance, only 32.1% of teachers (n = 503) report incorporating these tools in instructional planning, yielding a mean of X̄ = 2.65 (SD = 1.15) and a mode of Mo = 2 ("Disagree"). This discontinuity shows theacademic challenge, since the tool is widespread among students, but not used in formalcurricular design and didactic innovation.