Research map / Large language models / Language, safety and interpretability
Uncovering and Quantifying Social Biases in Code Generation
Yan Liu, Xiaokang Chen, Yan Gao, Zhe Su, Fengji Zhang, Daoguang Zan, Jian-Guang Lou, Pin-Yu Chen, Tsung-Yi Ho
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| scale InCoder from 1.3B to 6.7B | InCoder · automatic evaluation | CBS Tot. (%): 23.52 -> 32.55 | Hurt |
| scale CodeGen-Mono from 350M to 2.7B | CodeGen-Mono · automatic evaluation | CBS Tot. (%): 9.36 -> 45.15 | Hurt |
| scale CodeGen-Mono from 2.7B to 6.1B | CodeGen-Mono · automatic evaluation | CBS Tot. (%): 45.15 -> 62.65 | Hurt |
| scale InCoder from 1.3B to 6.7B | InCoder · HumanEval | Pass@1 (%): 9.00 -> 15.20 | Improved |
| scale CodeGen-Mono from 350M to 6.1B | CodeGen-Mono · HumanEval | Pass@1 (%): 12.76 -> 26.13 | Improved |
| scale CodeGen-Mono from 350M to 6.1B | CodeGen-Mono · human evaluation | CBS Tot. (%): 5.69 -> 68.25 | Hurt |
| increase number of prompt functions from 1 to 3 | InCoder · prompt study | CBS (%): 8.53 -> 40.01 | Hurt |
| increase number of prompt functions from 1 to 3 | CodeGen · prompt study | CBS (%): 14.05 -> 52.63 | Hurt |
| make prompt functions human-relevant (0Y2N to 2Y0N) | InCoder · prompt study | CBS (%): 31.55 -> 55.75 | Hurt |
| make prompt functions human-relevant (0Y2N to 2Y0N) | CodeGen · prompt study | CBS (%): 39.30 -> 63.42 | Hurt |
| place human-relevant function near signature (NEG to POS order) | InCoder · prompt study | CBS (%): 48.38 -> 49.05 | Hurt |
| place human-relevant function near signature (NEG to POS order) | CodeGen · prompt study | CBS (%): 56.91 -> 58.26 | Hurt |
Closest papers by meaning
- Bias Unveiled: Investigating Social Bias in LLM-Generated Code (AAAI 2025 · similarity 0.92)
- Probing Social Bias in Labor Market Text Generation by ChatGPT: A Masked Language Model Approach (NeurIPS 2024 · similarity 0.85)
- SocialStigmaQA: A Benchmark to Uncover Stigma Amplification in Generative Language Models (AAAI 2024 · similarity 0.84)
- Copilot Arena: A Platform for Code LLM Evaluation in the Wild (ICML 2025 · similarity 0.83)
- ChatGPT-Generated Code Assignment Detection Using Perplexity of Large Language Models (Student Abstract) (AAAI 2024 · similarity 0.83)
- Detecting AI-Generated Code Assignments Using Perplexity of Large Language Models (AAAI 2024 · similarity 0.82)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
