When Policies Change Probabilities: Modular Decision-Making for LLM Code Review
Source
Rasvik Kudum, Max Corbett, Hitansh Paliwal, Romaisa Fatima, Thomas Jiralerspong, Sneheel Sarangi
Author
Rasvik Kudum, Max Corbett, Hitansh Paliwal, Romaisa Fatima, Thomas Jiralerspong, Sneheel Sarangi
Date
Key takeaways · AI-distilled
Across 15,792 responses on 720 patches, telling a reviewer the cost of mistakes changes its reported failure probabilities by 13.6-16.9 points on identical evidence - the policy corrupts the estimate it is supposed to consume.
Under a 10:1 false-accept cost prompt, every one of the four deployed reviewers returns actions WORSE than rejecting all patches outright.
Applying the same high-cost rule to probabilities elicited WITHOUT policy information reduces loss for all four systems - proof the elicitation step itself, not just the decision rule, contributes the excess loss.
Their modular pipeline (policy-blind risk elicitation + independent monitor score + costs applied in code) cuts mean loss by .073 per issue at equal costs while accepting 58-68% of patches.
Practical rule: evaluate risk, outside evidence and action separately - never let the downstream policy leak into the probability prompt.
Terms in this piece · Glossary
LLM — A large language model — the neural network behind tools like Claude and ChatGPT, trained on huge amounts of text to predict what comes next.
embedding — A list of numbers representing a piece of text's meaning, so that similar meanings end up numerically close and can be searched.
calibration — How well a model's confidence matches reality — a calibrated model saying "90% sure" is right about 90% of the time.
Why it matters
embeddingA list of numbers representing a piece of text's meaning, so that similar meanings end up numerically close and can be searched.Full definition → cost-sensitive approval policies directly in a risk-estimation prompt can corrupt the reported probabilities themselves, sometimes making decisions worse than a blanket rejection under high-cost settings; separating probability elicitation from policy and using an independent monitor score improves calibrationHow well a model's confidence matches reality — a calibrated model saying "90% sure" is right about 90% of the time.Full definition → and reduces loss.
Key quotes
“A probability should depend on evidence; costs should determine the action taken from it.”
“For every reviewer, the actions returned under the high-cost prompt are worse than rejecting all patches.”
“Downstream policy can therefore change the probability it is meant to use, motivating separate evaluation of risk, outside evidence, and action.”
