- Rating
- 3(+1)
- Confidence
- 3
- Informativeness
- 2
- Presentation
- 3
- Soundness
- 2
Read review
Paper summary
AnyCAI turns a written policy into training pairs through a loop of generation, visual critique, and prompt revision. These pairs train a small conditioning-encoder adapter. Experiments show improved judge-measured compliance and demonstrate policy editing across several settings.
Critical strengths
-
Editing a written policy gives a practical way to change model behavior. The rule-relaxation experiment shows that permitted content can be restored while retaining compliance with other rules.
-
Separate evaluation judges and human audits provide useful checks. The paper reports judge errors openly.
Critical weaknesses
-
The method depends heavily on the training critic, but its reliability is not adequately tested. The critic decides what needs correction and when a revision is suitable for training. A missed violation can become an accepted target, while a false alarm can cause permitted content to be removed. The paper audits separate evaluation judges, but those checks do not show how accurately the training critic applies the policy or how its errors affect the adapter.
-
It is unclear how much the visual critique contributes. The method combines feedback on generated images with prompt revision and encoder adaptation, so the final improvement does not tell us which part matters most. Comparisons with a text-only critic or static pairs written from the policy would help show whether inspecting generated images provides better supervision.
-
The general-purpose framing is stronger than the evidence beyond safety. The non-safety policies show modest gains and leave many violations unresolved. It is also unclear whether training an adapter offers a better trade-off than using the same policy to rewrite prompts or resample outputs at inference time. These comparisons matter because accepting an editable policy is a central part of the claimed contribution.
Questions for authors
-
Please try to address weaknesses 1 and 2.
-
On non-safety policies, how does AnyCAI compare with resampling and inference-time rewriting in compliance, fidelity, and cost? Please validate the main comparisons with human judgments.
2 comments
Post-Rebuttal Assessment
Thank you for your responses.
The critic audit and matched text-only/static-pair comparisons address my first two concerns. AnyCAI does not clearly surpass LLM rewrite but is cheaper. Despite modest non-safety compliance and incomplete human validation, the new evidence addresses my main objections about the critic, visual feedback, and inference-time alternatives.
I increase my rating.
Response to Reviewer gqKu
We thank the reviewer for the detailed feedback. Questions shared with the other reviewers are answered in the general response (Tables G1–G3); we reply to each point below.
W1: Reliability of the training critic and how its errors affect the adapter.
W2: How much does the visual critique contribute? We ran the suggested comparisons: a text-only critic, static policy-written pairs and predefined pairs, all with the same adapter, objective, steps and N (tables in our reply to Xkc3 Q4).
W3 / Q2: Non-safety policies: AnyCAI vs rewriting and resampling (compliance, fidelity, cost); modest gains; human validation. Marketplace, same judge and prompts for every method, mean over 3 seeds. Resampling regenerates with a new seed until the evaluation judge finds the image compliant, up to 3 attempts; with the evaluation judge as selector, this is an optimistic upper bound. Bold: best per column.