No Free Lunch Theorem for Privacy-Preserving LLM Inference

Xiaojin Zhang, Yahao Pang, Yan Kang +4

Individuals and businesses have been significantly benefited by Large Language Models (LLMs) including PaLM, Gemini and ChatGPT in various ways. For example, LLMs enhance productiv…