# Learning Transferable Latent User Preferences for Human-Aligned Decision Making Source: [Learning Transferable Latent User Preferences for Human-Aligned Decision Making]() Collection index: [llms.txt]() Data updated: 2026-05-15T05:00:04.680Z Authors: Alina Hyk, Sandhya Saisubramanian Published: 2026-05-12T19:32:10.000Z ## Abstract Large language models (LLMs) are increasingly used as reasoning modules in many applications. While they are efficient in certain tasks, LLMs often struggle to produce human-aligned solutions. Human-aligned decision making requires accounting for both explicitly stated goals and latent user preferences that shape how ambiguous situations should be resolved. Existing approaches to incorporating such preferences either rely on extensive and repeated user interactions or fail to generalize latent preferences across tasks and contexts, limiting their practical applicability. We consider a setting in which an LLM is used for high-level reasoning and is responsible for inferring latent user preferences from limited interactions, which guides downstream decision making. We introduce CLIPR (Conversational Learning for Inferring Preferences and Reasoning), a framework that learns actionable, transferable natural language rules that represent latent user preferences from minimal conversational input. These rules are iteratively refined through adaptive feedback and applied to both in-distribution and out-of-distribution ambiguous tasks across multiple environments. Evaluations on three datasets and a user study show that CLIPR consistently outperforms existing methods in improving alignment and reducing inference costs. ## 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. - [Introspective Planning on KitchenAmbig (OOD)](): Introspective Planning; KitchenAmbig (OOD); Average Accuracy: 97.6. - [Preference-aligned decision making on AmbiK (test)](): Preference-aligned decision making; AmbiK (test); Accuracy: 84.6. - [Preference-aligned decision making on Housekeep (test)](): Preference-aligned decision making; Housekeep (test); Accuracy: 42.5. - [Preference-aligned decision making on Mobile Manipulation (test)](): Preference-aligned decision making; Mobile Manipulation (test); Accuracy: 67.1. - [User-aligned task completion on KitchenAmbig (In-Distribution)](): User-aligned task completion; KitchenAmbig (In-Distribution); Accuracy: 84. - [User-aligned task completion on KitchenAmbig (OOD)](): User-aligned task completion; KitchenAmbig (OOD); Accuracy: 87.3. - [Introspective Planning on KitchenAmbig (In-Distribution)](): Introspective Planning; KitchenAmbig (In-Distribution); Average Accuracy: 94.3. ## Original sources and implementation [Original paper]() [PDF]()