Pith. sign in

REVIEW 12 cited by

Measuring Implicit Bias in Explicitly Unbiased Large Language Models

Not yet reviewed by Pith; the record is open.

This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.

SPECIMEN: schema-true, not a live event

T0 review · schema-true

One-sentence machine reading of the paper's core claim.

pith:XXXXXXXX · record.json · timestamp

arxiv 2402.04105 v2 pith:LB5JQMHR submitted 2024-02-06 cs.CY cs.CL

Measuring Implicit Bias in Explicitly Unbiased Large Language Models

classification cs.CY cs.CL
keywords biasimplicitbiasesmeasuresdecisionlanguagellmsmeasuring
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved
0 comments
read the original abstract

Large language models (LLMs) can pass explicit social bias tests but still harbor implicit biases, similar to humans who endorse egalitarian beliefs yet exhibit subtle biases. Measuring such implicit biases can be a challenge: as LLMs become increasingly proprietary, it may not be possible to access their embeddings and apply existing bias measures; furthermore, implicit biases are primarily a concern if they affect the actual decisions that these systems make. We address both challenges by introducing two new measures of bias: LLM Implicit Bias, a prompt-based method for revealing implicit bias; and LLM Decision Bias, a strategy to detect subtle discrimination in decision-making tasks. Both measures are based on psychological research: LLM Implicit Bias adapts the Implicit Association Test, widely used to study the automatic associations between concepts held in human minds; and LLM Decision Bias operationalizes psychological results indicating that relative evaluations between two candidates, not absolute evaluations assessing each independently, are more diagnostic of implicit biases. Using these measures, we found pervasive stereotype biases mirroring those in society in 8 value-aligned models across 4 social categories (race, gender, religion, health) in 21 stereotypes (such as race and criminality, race and weapons, gender and science, age and negativity). Our prompt-based LLM Implicit Bias measure correlates with existing language model embedding-based bias methods, but better predicts downstream behaviors measured by LLM Decision Bias. These new prompt-based measures draw from psychology's long history of research into measuring stereotype biases based on purely observable behavior; they expose nuanced biases in proprietary value-aligned LLMs that appear unbiased according to standard benchmarks.

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Forward citations

Cited by 12 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. StereoTales: A Multilingual Framework for Open-Ended Stereotype Discovery in LLMs

    cs.CY 2026-05 unverdicted novelty 7.0

    StereoTales shows that LLMs produce harmful, culturally adapted stereotypes in open-ended multilingual stories, with patterns consistent across providers and aligned human-LLM harm judgments.

  2. StereoTales: A Multilingual Framework for Open-Ended Stereotype Discovery in LLMs

    cs.CY 2026-05 accept novelty 7.0

    StereoTales shows that all tested LLMs emit harmful stereotypes in open-ended stories, with associations adapting to prompt language and targeting locally salient groups rather than transferring uniformly across languages.

  3. Evalet: Evaluating Large Language Models through Functional Fragmentation

    cs.HC 2025-09 conditional novelty 7.0

    Evalet applies functional fragmentation to deliver fragment-level qualitative analysis of LLM evaluations, with a user study showing 48% more misalignment detections than holistic scoring.

  4. Artificial Intolerance: Stigmatizing Language in Clinical Documentation Skews Large Language Model Decision-Making

    cs.CL 2026-05 unverdicted novelty 6.0

    Frontier LLMs exhibit bias from stigmatizing language in clinical vignettes across four conditions, skewing decisions toward less aggressive management, with limited mitigation from Chain-of-Thought or self-debiasing prompts.

  5. Dialect vs Demographics: Quantifying LLM Bias from Implicit Linguistic Signals vs. Explicit User Profiles

    cs.CY 2026-04 unverdicted novelty 6.0

    Explicit demographic statements trigger higher refusal rates and lower semantic similarity in LLMs than implicit dialect cues, which reduce refusals but also reduce content sanitization.

  6. Aligned Agents, Biased Swarm: Measuring Bias Amplification in Multi-Agent Systems

    cs.MA 2026-04 unverdicted novelty 6.0

    Multi-agent systems amplify minor stochastic biases into systemic polarization via echo-chamber effects in structured workflows, even with neutral agents.

  7. Understanding and Mitigating Bias Inheritance in LLM-based Data Augmentation on Downstream Tasks

    cs.LG 2025-02 unverdicted novelty 6.0

    Empirical study across 10 tasks showing bias inheritance from LLM-augmented data harms related downstream performance, with three misalignment factors and three mitigation strategies identified.

  8. Evaluating Stochastic Collapse and Implicit Bias in Multimodal Large Language Models

    cs.CL 2026-06 unverdicted novelty 5.0

    MLLMs show stochastic collapse with top-1 probabilities up to 97% and low randomness indices when choosing among equivalent options.

  9. Reducing Political Manipulation with Consistency Training

    cs.CL 2026-05 unverdicted novelty 5.0

    Introduces Political Consistency Training (PCT) with sentiment and helpfulness consistency objectives to reduce covert political bias in LLMs while preserving helpfulness.

  10. Reducing Political Manipulation with Consistency Training

    cs.CL 2026-05 unverdicted novelty 5.0

    PCT is a reinforcement learning approach that trains LLMs for symmetric sentiment and helpfulness across paired opposing political prompts, reducing covert bias while preserving general performance.

  11. First, Do No Harm (With LLMs): Mitigating Racial Bias via Agentic Workflows

    cs.CY 2026-04 unverdicted novelty 4.0

    All five tested LLMs deviated from US race-stratified disease distributions in synthetic case generation, while retrieval-based agentic workflows improved mean p-value by 0.0348, median p-value by 0.1166, and mean dif...

  12. Inertia in Moral and Value Judgments of Large Language Models

    cs.CL 2024-08 unverdicted novelty 4.0

    LLMs exhibit persistent inertia in value orientations, with harm avoidance and fairness remaining skewed across persona prompts.