mlsanityData to 14 Oct 2026

Implicit Bias-Like Patterns in Reasoning Models

arXiv 2503.11572 ↗First posted 14 Mar 2025cs.CY cs.AIcited by 3

On the map: LLM reasoning → Reasoning benchmarks and evaluation

Abstract (arXiv)

Implicit biases refer to automatic mental processes that shape perceptions, judgments, and behaviors. Previous research on "implicit bias" in LLMs focused primarily on outputs rather than the processes underlying the outputs. We present the Reasoning Model Implicit Association Test (RM-IAT) to study implicit bias-like processing in reasoning models, LLMs that use step-by-step reasoning to solve complex tasks. Using RM-IAT, we find that reasoning models like o3-mini, DeepSeek-R1, gpt-oss-20b, and Qwen-3 8B consistently expend more reasoning tokens on association-incompatible tasks than association-compatible tasks, suggesting greater computational effort when processing counter-stereotypical information. Conversely, Claude 3.7 Sonnet exhibited reversed patterns, which thematic analysis associated with its unique internal focus on reasoning about bias and stereotypes. These findings demonstrate that reasoning models exhibit distinct implicit bias-like patterns and that these patterns vary significantly depending on the models' internal reasoning content.

Key facts. Cited by 3 papers in the map (data through 14 Oct 2026). Placed in Reasoning benchmarks and evaluation, one of 17 directions in LLM reasoning. Builds on 9 of its 62 references that are in the map. First posted on arXiv 14 Mar 2025.

Builds on

9 of 62 references.

References · 9

  1. 2022Training language models to follow instructions with human feedbackcited by 15,971
  2. 2022Chain-of-Thought Prompting Elicits Reasoning in Large Language Modelscited by 12,459
  3. 2023Direct Preference Optimization: Your Language Model is Secretly a Reward Modelcited by 8,065
  4. 2025DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learningcited by 6,330
  5. 2019Fine-Tuning Language Models from Human Preferencescited by 2,150
  6. 2016Semantics derived automatically from language corpora contain human-like biasescited by 901
  7. 2022Red Teaming Language Models to Reduce Harms: Methods, Scaling Behaviors, and Lessons Learnedcited by 729
  8. 2021Persistent Anti-Muslim Bias in Large Language Modelscited by 377
  9. 2024AIR-Bench 2024: A Safety Benchmark Based on Risk Categories from Regulations and Policiescited by 74

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2025: 2; 2026: 1.

3 papers.

  1. 2026Does Reasoning Make Search More Fair? Comparing Fairness in Reasoning and Non-Reasoning Rerankerscited by 2
  2. 2025Evaluating Implicit Biases in LLM Reasoning through Logic Grid Puzzlescited by 1
  3. 2025AI reasoning effort predicts human decision time in content moderationcited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2026Inference-Time Reasoning Selectively Reduces Implicit Social Bias in Large Language Models
  2. 2025Implicit Bias in LLMs: A Survey
  3. 2026BiasTrace: Linking Reasoning Behaviours to Biased Outputs in LLMs
  4. 2025Does Reasoning Introduce Bias? A Study of Social Bias Evaluation and Mitigation in LLM Reasoning
  5. 2024Do LLMs Really Think Step-by-step In Implicit Reasoning?
  6. 2025Beyond Words: A Latent Memory Approach to Internal Reasoning in LLMs
  7. 2026How Implicit Bias Accumulates and Propagates in LLM Long-term Memory
  8. 2025Implicit Reasoning in Large Language Models: A Comprehensive Survey