Aligning Language Models to User Opinions

An important aspect of developing LLMs that interact with humans is to align models' behavior to their users. It is possible to prompt an LLM into behaving as a certain persona, especially a user group or ideological persona the model captured during its pertaining stage. But, how to best align an LLM with a specific user and not a demographic or ideological group remains an open question. Mining public opinion surveys (by Pew Research), we find that the opinions of a user and their demographics and ideologies are not mutual predictors. We use this insight to align LLMs by modeling both user opinions as well as user demographics and ideology, achieving up to 7 points accuracy gains in predicting public opinions from survey questions across a broad set of topics. In addition to the typical approach of prompting LLMs with demographics and ideology, we discover that utilizing the most relevant past opinions from individual users enables the model to predict user opinions more accurately.

Self-Refine: Iterative Refinement with Self-Feedback

Like people, LLMs do not always generate the best text for a given generation problem on their first try (e.g., summaries, answers, explanations). Just as people then refine their text, we introduce SELF-REFINE, a framework for similarly improving initial outputs from LLMs through iterative feedback and refinement. The main idea is to generate an output using an LLM, then allow the same model to provide multi-aspect feedback for its own output; finally, the same model refines its previously generated output given its own feedback. Unlike earlier work, our iterative refinement framework does not require supervised training data or reinforcement learning, and works with a single LLM. We experiment with 7 diverse tasks, ranging from review rewriting to math reasoning, demonstrating that our approach outperforms direct generation. In all tasks, outputs generated with SELF-REFINE are preferred by humans and by automated metrics over those generated directly with GPT-3.5 and GPT-4, improving on average by absolute 20% across tasks.

Controlling Bias Exposure for Fair Interpretable Predictions

Recent work on reducing bias in NLP models usually focuses on protecting or isolating information related to a sensitive attribute (like gender or race). However, when sensitive information is semantically entangled with the task information of the input, e.g., the gender information is predictive for a profession, a fair trade-off between task performance and bias mitigation is difficult to achieve. Existing approaches perform this trade-off by eliminating bias information from the latent space, lacking control over how much bias is necessarily required to be removed. We argue that a favorable debiasing method should use sensitive information 'fairly' rather than blindly eliminating it (Caliskan et al., 2017; Sun et al., 2019). In this work, we provide a novel debiasing algorithm by adjusting the predictive model's belief to (1) ignore the sensitive information if it is not useful for the task; (2) use sensitive information minimally as necessary for the prediction (while also incurring a penalty). Experimental results on two text classification tasks (influenced by gender) and an open-ended generation task (influenced by race) indicate that our model achieves a desirable trade-off between debiasing and task performance along with producing debiased rationales as evidence.

InterFair: Debiasing with Natural Language Feedback for Fair Interpretable Predictions

Debiasing methods in NLP models traditionally focus on isolating information related to a sensitive attribute (like gender or race). We instead argue that a favorable debiasing method should use sensitive information 'fairly,' with explanations, rather than blindly eliminating it. This fair balance is often subjective and can be challenging to achieve algorithmically. We show that an interactive setup with users enabled to provide feedback can achieve a better and fair balance between task performance and bias mitigation, supported by faithful explanations.

On Faithfulness and Coherence of Language Explanations for Recommendation Systems

Reviews contain rich information about product characteristics and user interests and thus are commonly used to boost recommender system performance. Specifically, previous work show that jointly learning to perform review generation improves rating prediction performance. Meanwhile, these model-produced reviews serve as recommendation explanations, providing the user with insights on predicted ratings. However, while existing models could generate fluent, human-like reviews, it is unclear to what degree the reviews fully uncover the rationale behind the jointly predicted rating. In this work, we perform a series of evaluations that probes state-of-the-art models and their review generation component. We show that the generated explanations are brittle and need further evaluation before being taken as literal rationales for the estimated ratings.

Achieving Conversational Goals with Unsupervised Post-hoc Knowledge Injection

A limitation of current neural dialog models is that they tend to suffer from a lack of specificity and informativeness in generated responses, primarily due to dependence on training data that covers a limited variety of scenarios and conveys limited knowledge. One way to alleviate this issue is to extract relevant knowledge from external sources at decoding time and incorporate it into the dialog response. In this paper, we propose a post-hoc knowledge-injection technique where we first retrieve a diverse set of relevant knowledge snippets conditioned on both the dialog history and an initial response from an existing dialog model. We construct multiple candidate responses, individually injecting each retrieved snippet into the initial response using a gradient-based decoding method, and then select the final response with an unsupervised ranking step. Our experiments in goal-oriented and knowledge-grounded dialog settings demonstrate that human annotators judge the outputs from the proposed method to be more engaging and informative compared to responses from prior dialog systems. We further show that knowledge-augmentation promotes success in achieving conversational goals in both experimental settings.

Self-Supervised Bot Play for Conversational Recommendation with Justifications

Conversational recommender systems offer the promise of interactive, engaging ways for users to find items they enjoy. We seek to improve conversational recommendation via three dimensions: 1) We aim to mimic a common mode of human interaction for recommendation: experts justify their suggestions, a seeker explains why they don't like the item, and both parties iterate through the dialog to find a suitable item. 2) We leverage ideas from conversational critiquing to allow users to flexibly interact with natural language justifications by critiquing subjective aspects. 3) We adapt conversational recommendation to a wider range of domains where crowd-sourced ground truth dialogs are not available. We develop a new two-part framework for training conversational recommender systems. First, we train a recommender system to jointly suggest items and justify its reasoning with subjective aspects. We then fine-tune this model to incorporate iterative user feedback via self-supervised bot-play. Experiments on three real-world datasets demonstrate that our system can be applied to different recommendation models across diverse domains to achieve superior performance in conversational recommendation compared to state-of-the-art methods. We also evaluate our model on human users, showing that systems trained under our framework provide more useful, helpful, and knowledgeable recommendations in warm- and cold-start settings.

Detect and Perturb: Neutral Rewriting of Biased and Sensitive Text via Gradient-based Decoding

Written language carries explicit and implicit biases that can distract from meaningful signals. For example, letters of reference may describe male and female candidates differently, or their writing style may indirectly reveal demographic characteristics. At best, such biases distract from the meaningful content of the text; at worst they can lead to unfair outcomes. We investigate the challenge of re-generating input sentences to 'neutralize' sensitive attributes while maintaining the semantic meaning of the original text (e.g. is the candidate qualified?). We propose a gradient-based rewriting framework, Detect and Perturb to Neutralize (DEPEN), that first detects sensitive components and masks them for regeneration, then perturbs the generation model at decoding time under a neutralizing constraint that pushes the (predicted) distribution of sensitive attributes towards a uniform distribution. Our experiments in two different scenarios show that DEPEN can regenerate fluent alternatives that are neutral in the sensitive attribute while maintaining the semantics of other attributes.

Rationale-Inspired Natural Language Explanations with Commonsense

Explainable machine learning models primarily justify predicted labels using either extractive rationales (i.e., subsets of input features) or free-text natural language explanations (NLEs) as abstractive justifications. While NLEs can be more comprehensive than extractive rationales, machine-generated NLEs have been shown to sometimes lack commonsense knowledge. Here, we show that commonsense knowledge can act as a bridge between extractive rationales and NLEs, rendering both types of explanations better. More precisely, we introduce a unified framework, called RExC (Rationale-Inspired Explanations with Commonsense), that (1) extracts rationales as a set of features responsible for machine predictions, (2) expands the extractive rationales using available commonsense resources, and (3) uses the expanded knowledge to generate natural language explanations. Our framework surpasses by a large margin the previous state-of-the-art in generating NLEs across five tasks in both natural language processing and vision-language understanding, with human annotators consistently rating the explanations generated by RExC to be more comprehensive, grounded in commonsense, and overall preferred compared to previous state-of-the-art models. Moreover, our work shows that commonsense-grounded explanations can enhance both task performance and rationales extraction capabilities.

Unsupervised Enrichment of Persona-grounded Dialog with Background Stories

Humans often refer to personal narratives, life experiences, and events to make a conversation more engaging and rich. While persona-grounded dialog models are able to generate responses that follow a given persona, they often miss out on stating detailed experiences or events related to a persona, often leaving conversations shallow and dull. In this work, we equip dialog models with 'background stories' related to a persona by leveraging fictional narratives from existing story datasets (e.g. ROCStories). Since current dialog datasets do not contain such narratives as responses, we perform an unsupervised adaptation of a retrieved story for generating a dialog response using a gradient-based rewriting technique. Our proposed method encourages the generated response to be fluent (i.e., highly likely) with the dialog history, minimally different from the retrieved story to preserve event ordering and consistent with the original persona. We demonstrate that our method can generate responses that are more diverse, and are rated more engaging and human-like by human evaluators, compared to outputs from existing dialog models.