GenCtrl -- A Formal Controllability Toolkit for Generative Models
This work addresses the need for rigorous controllability analysis in generative models, shifting focus from control attempts to understanding fundamental limits, which is significant for researchers and practitioners in AI.
The paper tackled the problem of assessing whether generative models are truly controllable by developing a theoretical framework and algorithm to estimate controllable sets with formal guarantees on error bounds, showing that model controllability is fragile and highly dependent on experimental settings.
As generative models become ubiquitous, there is a critical need for fine-grained control over the generation process. Yet, while controlled generation methods from prompting to fine-tuning proliferate, a fundamental question remains unanswered: are these models truly controllable in the first place? In this work, we provide a theoretical framework to formally answer this question. Framing human-model interaction as a control process, we propose a novel algorithm to estimate the controllable sets of models in a dialogue setting. Notably, we provide formal guarantees on the estimation error as a function of sample complexity: we derive probably-approximately correct bounds for controllable set estimates that are distribution-free, employ no assumptions except for output boundedness, and work for any black-box nonlinear control system (i.e., any generative model). We empirically demonstrate the theoretical framework on different tasks in controlling dialogue processes, for both language models and text-to-image generation. Our results show that model controllability is surprisingly fragile and highly dependent on the experimental setting. This highlights the need for rigorous controllability analysis, shifting the focus from simply attempting control to first understanding its fundamental limits.