Artificial intelligence in Latin American higher education: implementations, ethical challenges, and pedagogical effectiveness

Artificial intelligence is establishing itself as a catalyst for transformation in the regional university sector, generating growing yet uneven academic output. This research conducted a systematic review following the PRISMA methodology on applications of artificial intelligence in Latin American higher education. The results from the 421 studies obtained during the bibliometric stage indicate that research is geographically and institutionally concentrated in a limited set of approaches and practices. In this regard, a notable prevalence of studies on Machine Learning applications, as well as Natural Language Processing, was observed. From a practical standpoint, 30 studies were selected for qualitative analysis. These texts agreed that the implementation process of these technologies continues to face structural challenges. Notably, poor infrastructure conditions, as well as deficiencies in teacher training, were identified as the main obstacles to implementing these technologies. The analyzed studies also concurred on the inadequate treatment of algorithmic biases or data protection in application policies proposed by the literature. Consequently, a key recommendation of this research is the urgent need for studies aimed at evaluating short-term outcomes, as well as analyzing the long-term sustainability of such innovations.