Abstract:

In recent decades, higher education systems have been transforming due to rapid changes in digital tools, particularly the disruptive impact of generative AI. This technology has become a priority on international, national, and institutional agendas, leading to debates, tensions, and challenges that must be addressed. Los Andes University (Bogotá, Colombia) has taken steps to reflect on and implement strategies to achieve generative AI literacy for all members of its community. This text is part of a broader research project that seeks to answer the question: What evidence can be identified from the analysis of pedagogical practices in five courses across different faculties that introduce AI, contributing to the understanding and guidance of AI literacy processes at the curricular level at the University of the Andes? A multiple case study using concurrent mixed methods is being conducted. For each case (course), instruments such as semi-structured interviews with professors, graduate assistants, and teaching assistants; focus groups and student surveys; analysis of student-generated documents; and non-participant classroom observation are employed. The predefined categories are the four dimensions of digital literacy proposed by Kong et al. (2024): cognitive, metacognitive, affective, and social, which can be articulated with the transversal competencies proposed by the University of the Andes. The information is analyzed from a critical-hermeneutic perspective, providing both a general and a detailed understanding of the phenomenon and consolidating a valid and relevant discourse on the reality of the topic under study.

The preliminary results presented in this document are based on non-participant observation in five courses:
- (Medications: How Do They Work and How to Use Them?) Medicine,
- (Pedagogical Practice 1) Education,
- (Introduction to Mechanical Engineering) Engineering,
- (Humanitarianism and Power) Social Sciences, and
- (Quantum Chemistry and Spectroscopy) Natural Sciences.

The observations revealed that the conceptual presentation of AI is infrequent in these courses, with a greater emphasis on its application. Several courses discuss prompt quality, strategies to improve it, the advantages and disadvantages of various tools (such as ChatGPT and Copilot), and the risks of hallucinations. Some courses employ tools designed specifically for the course (e.g., the education bot or the mechanical engineering bot), and a critical analysis of the responses they generate is encouraged. Regarding the social dimension, the ethical implications of using these tools are discussed in some cases, especially in the medical course. In contrast, biases in these tools are rarely addressed in other classes. Finally, the observations do not reveal significant information about students' empowerment in the autonomous use of these tools.

Keywords:

Generative AI literacy, education, Generative AI in higher education, Pedagogical Practices with Generative AI.