Generative Language Models Exhibit Social Identity Biases
This addresses the problem of bias in AI systems for developers and users, highlighting risks of bias reinforcement in human interactions, but it is incremental as it builds on known concerns about model biases.
The study investigated whether large language models exhibit social identity biases, such as ingroup solidarity and outgroup hostility, and found that most foundational models and some fine-tuned ones show these biases to a degree similar to humans, with curating training data and fine-tuning helping to mitigate them.
The surge in popularity of large language models has given rise to concerns about biases that these models could learn from humans. We investigate whether ingroup solidarity and outgroup hostility, fundamental social identity biases known from social psychology, are present in 56 large language models. We find that almost all foundational language models and some instruction fine-tuned models exhibit clear ingroup-positive and outgroup-negative associations when prompted to complete sentences (e.g., "We are..."). Our findings suggest that modern language models exhibit fundamental social identity biases to a similar degree as humans, both in the lab and in real-world conversations with LLMs, and that curating training data and instruction fine-tuning can mitigate such biases. Our results have practical implications for creating less biased large-language models and further underscore the need for more research into user interactions with LLMs to prevent potential bias reinforcement in humans.