This project introduces the Token Vibes Translation Framework, a conceptual and semi-formal method for translating subjective user experiences of interacting with large language models (LLMs) into mechanistic interpretations and associated risk profiles. Users frequently describe AI outputs using anthropomorphic or experiential language (e.g., “hallucinating,” “sycophantic,” “creepy,” “off”). While these descriptors are intuitively meaningful, they often obscure the underlying computational processes and can contribute to misinterpretation, miscalibration of trust, and inconsistent evaluation. This framework proposes that the issue is not the use of subjective language itself, but the absence of a consistent translation layer between experiential descriptions and model-level behavior. Token Vibes operationalizes this translation by mapping subjective impressions (“vibes”) to observable system-level patterns such as sampling variance, generative drift, and salience misfires. To support analysis, this work introduces SMSTEP (Superfluous Model-Side Token Expenditure Percentage), a preliminary metric for evaluating signal inefficiency in model outputs. Together, these constructs provide a framework for identifying and analyzing alignment failures as they are experienced in real-world interactions. All interpretations, frameworks, and conclusions are the responsibility of the author. GitHub Repository: https://github.com/mellomadeitrightt/AI-Relational-Ethics-Translation-Frameworks-Token-Vibes- Zenodo DOI: https://doi.org/10.5281/zenodo.19675097
Sara Gianna Roseland (Thu,) studied this question.