Abstract:

The rise of generative artificial intelligence (GenAI) has added new layers of complexity to today’s educational landscape. While these technologies present extraordinary opportunities to enrich learning experiences, they also introduce significant challenges for educators. Ensuring academic integrity, cultivating original and critical thinking, and adapting teaching strategies in an era of instant information access require a reexamination of traditional methods. As AI continues to advance, educators must strike a careful balance between embracing innovation and safeguarding intellectual development.

We have encountered these needs and challenges when teaching in a Degree in Computer Science and Engineering at University of Castilla-La Mancha, Spain. Traditional approaches to learning checks and evaluation, such as assigning exercises to be solved at home and later graded, are no longer adequate. Students can now use GenAI systems to generate solutions in a fraudulent manner, much like copying from classmates in the past. Consequently, educators cannot rely on conventional formats for class exercises, assignments, and exams, as they fail to provide valid assessments of students’ learning and performance.

As a solution, we present our experience in revising how we teach, promote students' learning and assess it in the current technological context. We have used and are using the strategies in various courses covering different topics, such as programming, information systems, and software engineering. We have adjusted the format of class exercises, assignments and exams around three key aspects. First, activities whose outcome is graded (e.g., a practical programming exam) should not be performed with connection to the Internet, with students’ computers, or as home assignment (unless in the scope of the third aspect below). Homework can still be assigned but not graded; instead, it serves as preparation for subsequent sessions or as a prerequisite for passing. Second, class exercises (e.g., true/false statements about certain software testing techniques) should more largely rely on paper-based formats, possibly combined with information on a board. This prevents students from simply querying GenAI systems for answers, whether through text or image input. Third, GenAI usage must be integrated in some way so that students learn how to effectively and responsibly use it (e.g., to create a database schema), while also gaining awareness of how these systems are developed and work, and of their capabilities and limitations. When GenAI-based activities are graded, evaluation should emphasize the creation process and the student’s understanding of the outcome.

We argue that these formats for class exercises, assignments and exams both adapt to the realities of the GenAI era and integrate the technology into the learning environment. Ultimately, they contribute to an effective teaching-learning process aligned with current and future educational and professional needs. Last but not least, these strategies can be easily adapted to other subjects and disciplines.

Keywords:

Higher education, genAI, computer science.