Pseudointelligence: A Unifying Framework for Language Model Evaluation
It addresses the need for principled evaluation in AI as models surpass human benchmarks, though it is incremental in refining evaluation approaches.
The paper tackles the problem of evaluating language models by proposing pseudointelligence, a framework that incorporates the evaluator's perspective, and demonstrates its application to case studies and analysis of existing methods.
With large language models surpassing human performance on an increasing number of benchmarks, we must take a principled approach for targeted evaluation of model capabilities. Inspired by pseudorandomness, we propose pseudointelligence, which captures the maxim that "(perceived) intelligence lies in the eye of the beholder". That is, that claims of intelligence are meaningful only when their evaluator is taken into account. Concretely, we propose a complexity-theoretic framework of model evaluation cast as a dynamic interaction between a model and a learned evaluator. We demonstrate that this framework can be used to reason about two case studies in language model evaluation, as well as analyze existing evaluation methods.