CoS: Enhancing Personalization and Mitigating Bias with Context Steering

arXiv:2405.01768v1 Announce Type: new
Abstract: When querying a large language model (LLM), the context, i.e. personal, demographic, and cultural information specific to an end-user, can significantly shape the response of the LLM. For example, asking the model to explain Newton's second law with the context "I am a toddler" yields a different answer compared to the context "I am a physics professor." Proper usage of the context enables the LLM to generate personalized responses, whereas inappropriate contextual influence can lead …

abstract arxiv bias context cs.ai cs.cl example information language language model large language large language model law llm personalization type