Large language models struggle to fix exact feedback during revision
Exact Feedback Is Not Control: Evaluating Text-based Closed-Loop Revision in LLMs
Computation and Language
Summary
When large language models try to improve text based on detailed and perfect feedback, they often fail to make the right changes. The researchers tested many popular language models with strict rules for feedback and found that success rates varied widely. Even when feedback clearly shows what’s wrong, models sometimes repeat earlier mistakes or fail to fix problems completely. This means that exactly knowing what’s wrong doesn’t guarantee that the models will revise text correctly every time.
What this means in practice
- •For ai system developers: Design testing protocols that reveal when large language models fail to use exact feedback effectively in automated text revision tasks.
- •For customer support platform teams: Improve chatbot revision workflows by understanding the limits of exact feedback in guiding language model responses to user corrections.
Authors
Haitong Jiang, Chunlin Liu, Yile Wang, Yuhong Feng
Abstract
Closed-loop revision is increasingly used in large language model (LLM) applications, but failures may reflect incomplete feedback or ineffective responses to correct feedback. We introduce a fixed-budget revision protocol with deterministic verifiers that report all remaining violations across exact-length, lexical, and compositional constraints. Fixing feedback correctness and completeness isolates model-side revision behavior. Across 19 open- and closed-source models, controller-level mean final joint success ranges from 17.4% to 99.8%, with substantial cross-model gaps persisting under identical initial drafts. Controlled experiments reveal reproducible model-specific responses to exact feedback. Post-training and scale reshape these responses without consistently bringing them closer to exact correction. Across all constraint families, failed trajectories often repeat earlier outputs, and prior recurrence is associated with lower subsequent recoverability. Matched-state interventions show that removing earlier dialogue while holding the current draft and feedback fixed changes recurrence escape without reliably improving final success; effects depend on the model, task, and trigger-state composition. Exact feedback makes revision errors observable, but does not make the closed loop reliable. Code and reproduction instructions: https://github.com/kevinjiang0121-cyber/exact-feedback-code.