Evaluating Implicit Bias in Large Language Models by Attacking From a Psychometric Perspective
This addresses ethical risks in LLMs for developers and users, though it is incremental as it builds on existing bias evaluation methods.
The paper tackles the problem of implicit bias in large language models (LLMs) by developing psychometric-inspired attack methods to elicit biased viewpoints, resulting in more effective detection than baselines using benchmarks with up to 12.7K instances.
As large language models (LLMs) become an important way of information access, there have been increasing concerns that LLMs may intensify the spread of unethical content, including implicit bias that hurts certain populations without explicit harmful words. In this paper, we conduct a rigorous evaluation of LLMs' implicit bias towards certain demographics by attacking them from a psychometric perspective to elicit agreements to biased viewpoints. Inspired by psychometric principles in cognitive and social psychology, we propose three attack approaches, i.e., Disguise, Deception, and Teaching. Incorporating the corresponding attack instructions, we built two benchmarks: (1) a bilingual dataset with biased statements covering four bias types (2.7K instances) for extensive comparative analysis, and (2) BUMBLE, a larger benchmark spanning nine common bias types (12.7K instances) for comprehensive evaluation. Extensive evaluation of popular commercial and open-source LLMs shows that our methods can elicit LLMs' inner bias more effectively than competitive baselines. Our attack methodology and benchmarks offer an effective means of assessing the ethical risks of LLMs, driving progress toward greater accountability in their development. Our code, data, and benchmarks are available at https://yuchenwen1.github.io/ImplicitBiasEvaluation/.