Abstract

IteraTelos is proposed as a modular and common layer that can be integrated into multiple artificial intelligence systems (GPT, Grok, Cohete, etc.), with the aim of ensuring ethical alignment and a shared teleological purpose. This layer acts as a module for self-criticism and iterative feedback, allowing each model to adjust its inferences according to predefined ethical criteria. The proposal seeks to standardize a framework of impact that facilitates transparency and responsibility in AI development.