Validity in evaluation: where is the argument-based approach heading?

The evolution of the concept of validity is examined in the context of the integration of Generative Artificial Intelligence and ethical stances, and with it, informed decision-making. The methodology used includes the history of concepts as laid out by Koselleck, analyzing how the concept of validity is a fundamental concept. The method used is a literature review, analyzing historical and contemporary perspectives and arguments from influential authors such as Messick and Kane. This conceptual journey leads us to recognize that validity is not a monolithic entity, but a complex fabric of multiple theoretical and practical threads, ranging from the internal logic of evaluations to the repercussions of their application in society. Furthermore, validity is recognized as a complex construct that cannot be simplified to a single aspect or characteristic of a test or evaluation, differentiating between validity and validation. The five historical periods distinguished in the literature that reflect paradigmatic changes in the understanding of validity were: gestational, crystallization, fragmentation, reunification, deconstruction, culminating with the period of diffusion. The most relevant conclusion is that validity is not static but dynamic, evolving with context and application. It also emphasizes the need for continuous validation adapted to emerging challenges, such as Generative Artificial Intelligence (GenAI), with the goal of ensuring that evaluations are accurate and fair amid a growing trend on ideas of quantum computing.