The use of Artificial Intelligence (AI) in research and teaching has arisen several times in the past few meetings I have attended, either within the EECRG or within our Faggruppe, particularly in the relation to large-language models. On the one hand I have found these models as a useful additional tool to help with coding and data-analyses, poster and presentation design, and as an alternative to a search engine to summarise basic concepts.
One the other hand, hallucinated references, mixing up of coding languages, and some obvious factual inaccuracies taught me to be aware of the limitations of these technologies. Perhaps these problems will be ironed out in the future as the language models develop, but it’s clear one of the challenges we face in the university setting is for both students and researchers to find an optimal balance behind finding the potential value of such tools, without being overdependent and thinking uncritically about their outputs.
As I am sure many will agree, AI has presented major challenges in teaching over the past few years. Many of the assessment tools I used in my own course became less relevant as AI has become more widely used, and the rapidly developing technologies will likely mean we must constantly re-evaluate the types of assessments that are valuable for both teachers and students. The biggest questions to come from our discussions have been, how can we test for knowledge and understanding in the age of AI and in the longer term, how does the availability of AI alter the skills, knowledge and understanding our graduates need?
I believe this is a topic that will be raised at forthcoming BIOTEACH seminar in March. As will all the BIO-Thursday lunchtime seminars, this will definitely be worth attending.
Alistair Seddon
Group leader, Ecology and Evolution