On 22nd June, the session entitled "AI in teaching: Good practices, challenges and opportunities", organised by the Innovation Space of the School of Engineering and ICT Department was held at the Poblenou Campus to share experiences, challenges, reflections and concerns.

Participants in the session included the professor of the Department Davinia Hernández-Leo, who dynamizes the training on this subject at UPF and other institutions.

The agenda of the session consisted of an intervention by Professor Hernández-Leo, followed by a discussion moderated by the head of Teaching Innovation of the School, Professor Patricia Santos, who supported her intervention with this presentation.

Dr Hernández-Leo's talk:

In order to provide some context, Professor Hernández-Leo highlighted the impact of the use of tools such as Chat GPT in Higher Education and the fact that media have been talking about AI in general and its impact and use at this stage. In this respect, she commented on specific cases of countries where moratoriums have been considered and/or it has been discussed to return to the "past" in education, by using pencil and paper extensively in the classroom.

Professor Hernández-Leo also commented on the impact of these types of tools on plagiarism and the type of skills that can be worked on and improved with their use. She focused not only on the "negative" impact, but on how using them properly can positively impact the improvement of educational practices and, therefore, on the learning process. She emphasized that, now, it is more important, interesting and necessary than ever to reinforce the use of human intelligence in contrast to AI.

The limitations and origins of these tools were also discussed. Obviously, people with educational and technological backgrounds have significatively more resources and knowledge to perform much more refined work with these tools than other people who see more threats than opportunities. Thus, the crux of the matter is to refine and give the right "orders" to have more refined and rigorous results. That is not to say that these types of tools do not have limitations, but it is important to understand that these are also dependent on the use made of them.

More specifically, the professor pointed to what type of learning tasks these tools can be especially positive for. Among these, self-evaluation stands out: asking for feedback on a text at the grammar or coherence levels, or processes such as brainstorming. During her talk, Professor Hernández-Leo showed us an example of the use of GPT Chat that she herself carried out and highlighted, among other aspects, an error in the feedback by GPT Chat. This reinforces the idea that it is necessary to revise critically the feedback provided by these types of tools. Other examples of the use of GPT Chat and similar tools in education are to compare students' outputs, to compare different versions of outputs or texts, to detect improvements derived from changes in said versions, etc.

As challenges, she highlighted elements such as the professional responsibility that we all have as public workers in higher education and as potential generators of AI. Elements such as privacy, plagiarism, veracity and ethics (biases) are fundamental aspects to take into account while developing and using AI. She also pointed to the reflection on the irrefutable fact that AI technologies advance faster than the detectors of their use. Therefore, it is more efficient and enriching to make good use of these tools than to avoid them because it seems theoretically impossible.

Thus, Professor Hernández-Leo suggested considering criteria such as effectiveness, efficiency, authenticity and fun for the use of these tools in the learning process. She also said that it is important and necessary to rethink the evaluable outputs that we propose to our students, considering the processes and not only the products/results.

Regarding the discussion, we first summarised the feedback that was collected through the form provided to the attendees before the session. The form mentioned the use and type of interaction with AI (direct or indirect) that has occurred in different courses. Some examples were: to raise elements of reflection on the use and impact of AI in the profession, in the academic field, in its good use to obtain greater performance, in risks, etc.

The ethical issue appears in the behaviour/ attitude of students as an essential element since the use of AI tools is inevitable and even necessary as an educational resource. It is on the previous step that we should focus: how do we prepare students to behave ethically and at the same time extensively use AI to get the most out of it?

The main element of debate was undoubtedly evaluation, especially in numerous courses in which more one-to-one follow-up is not possible due to a lack of resources. At the same time, other types of evaluation methodologies were discussed, such as oral tests, handwritten tests, etc.

In future, we agreed that it is necessary to develop a guide and collect good practices at our School, a "lively resource" that will be built and will grow over time. It was also proposed to repeat the session at the beginning of the year.