# Intriguing properties of generative classifiers Source: [Intriguing properties of generative classifiers]() Collection index: [llms.txt]() Data updated: 2025-12-25T02:36:23.174Z Authors: Priyank Jaini, Kevin Clark, Robert Geirhos Published: 2023-09-28T18:19:40.000Z ## Abstract What is the best paradigm to recognize objects -- discriminative inference (fast but potentially prone to shortcut learning) or using a generative model (slow but potentially more robust)? We build on recent advances in generative modeling that turn text-to-image models into classifiers. This allows us to study their behavior and to compare them against discriminative models and human psychophysical data. We report four intriguing emergent properties of generative classifiers: they show a record-breaking human-like shape bias (99% for Imagen), near human-level out-of-distribution accuracy, state-of-the-art alignment with human classification errors, and they understand certain perceptual illusions. Our results indicate that while the current dominant paradigm for modeling human object recognition is discriminative inference, zero-shot generative models approximate human object recognition data surprisingly well. ## Recorded benchmark results A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage. - [Model-Human Comparison on model-vs-human (test)](): Model-Human Comparison; model-vs-human (test); Error Consistency: 0.31. ## Original sources and implementation [Original paper]() [PDF]()