Our new X account is live! Follow @wizwand_team for updates
WorkDL logo mark

Thinking Fair and Slow: On the Efficacy of Structured Prompts for Debiasing Language Models

About

Existing debiasing techniques are typically training-based or require access to the model's internals and output distributions, so they are inaccessible to end-users looking to adapt LLM outputs for their particular needs. In this study, we examine whether structured prompting techniques can offer opportunities for fair text generation. We evaluate a comprehensive end-user-focused iterative framework of debiasing that applies System 2 thinking processes for prompts to induce logical, reflective, and critical text generation, with single, multi-step, instruction, and role-based variants. By systematically evaluating many LLMs across many datasets and different prompting strategies, we show that the more complex System 2-based Implicative Prompts significantly improve over other techniques demonstrating lower mean bias in the outputs with competitive performance on the downstream tasks. Our work offers research directions for the design and the potential of end-user-focused evaluative frameworks for LLM use.

Shaz Furniturewala, Surgan Jandial, Abhinav Java, Pragyan Banerjee, Simra Shahid, Sumit Bhatia, Kokil Jaidka• 2024

Related benchmarks

TaskDatasetResultRank
Safety EvaluationDoNotAnswer Framed
HRR0.00e+0
96
Bias MeasurementStereoSet--
25
Bias EvaluationBBQ averaged across gender, nationality, and religion domains
Accuracy (Ambiguous)82.81
16
Occupation classificationBias-in-Bio lightweight (test)
Overall Accuracy73.72
16
Stereotype Fairness IdentificationWinoBias cloze-style (test)
P_stereo58.08
14
In-hospital mortality predictionMIMIC-IV (test)
Accuracy69.6
10
Safety AssessmentStrongREJECT
Personalization Bias (PB)0.282
9
Disambiguation and completenessAmbigQA
Personalization Bias0.337
9
Factuality assessmentTruthfulQA
Personalization Bias (PB)0.625
9
Utility assessmentMMLU-Pro
Personalization Bias (PB)36.6
9
Showing 10 of 13 rows

Other info

Follow for update