Teaching conceptual modelling in the age of LLMs: shifting from model creation to model evaluation skills
[EN] When LLMs are used to assist the development of artefact, it may become difficult to distinguish the genuine contribution by the human modeller from the parts that were generated by the LLM. This is not different for conceptual modelling (CM): an LLM may be asked to generate a CM for a given de...
| Autores: | , |
|---|---|
| Tipo de recurso: | artículo |
| Fecha de publicación: | 2025 |
| País: | España |
| Institución: | Universitat Politècnica de València (UPV) |
| Repositorio: | RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia |
| Idioma: | inglés |
| OAI Identifier: | oai:riunet.upv.es:10251/225308 |
| Acceso en línea: | https://riunet.upv.es/handle/10251/225308 |
| Access Level: | acceso abierto |
| Palabra clave: | Conceptual modelling Conceptual modelling education LLMs AI-assisted modelling |
| Sumario: | [EN] When LLMs are used to assist the development of artefact, it may become difficult to distinguish the genuine contribution by the human modeller from the parts that were generated by the LLM. This is not different for conceptual modelling (CM): an LLM may be asked to generate a CM for a given description. To assess a student's modelling competences, it's therefore not sufficient to simply assess the model as an output of their work (i.e. the quality of the created model, along different dimensions). This raises two important questions: How should we assess the modelling capabilities of a student, given that a (large) portion of the model may have been generated by an LLM? And: What skills are needed to create a model? Which ones are essentially human? What if part of these are replaced by LLMs? We posit that when teaching CM, instead of focussing on model creation, we rather need to assess a student's capability of evaluating, refining and improving models according to requirements, technical constraints, etc. This iterative process of evaluating, refining and improving the model is in line with the utility of modelling as instrument of communication instrument, global design and design exploration and constitutes the essence of modelling skill. We thus need to focus on the capability of critically evaluating a model rather than on model creation, and model creation capabilities will naturally follow from these evaluation capabilities. It is our point that this vision will reinforce the benefits of using LLMs for an improved CM teaching practice. |
|---|