Briefing
The geneticist Lluís Montoliu appeared in Logroño on 29 October 2026 and warned against using generative artificial intelligence systems’ outputs in academic work without checking them. Montoliu, deputy director at the Centro Nacional de Biotecnología (CNB) of the Consejo Superior de Investigaciones Científicas (CSIC) and a member of the CIBERERnetwork at the Instituto de Salud Carlos III, spoke as part of university events at the Universidad de La Rioja on the occasion of the “Día de Bienvenida del Doctorandoannounced for that day, as well as related talks on the responsible use of AI in research.
He explicitly recommended that doctoral students test model responses with questions from their own field: only those who truly know a subject can reliably identify and substantiate inaccurate or invented information. Montoliu stressed that generative models can make routine tasks easier, but can also produce plausible-sounding false data that may slip unnoticed into texts, data interpretations or assessments.
He also argued for making training a required part of university courses: technical instruction alone is not enough; teaching is needed on checking sources, documenting uncertainty and institutional rules for disclosing the limitations of models. Supervisors and examination boards should be involved in these processes to ensure quality standards are maintained.
According to current research, there is no scientific evidence that expert questioning of models reliably prevents misinformation across all disciplines; Montoliu therefore presented his recommendations as a reasoned practice that should be continually evaluated. His talk is part of a wider debate in Spain about integrating artificial intelligence into research and teaching, which has recently been discussed at several public events and in media reports.
Editor: Mira Cadena (Article created with AI assistance).