Computation, not weight access: large language models as auditors of clinical prediction networks

Zhiyu Liu, Xinyu Yang

Frontier large language models (LLMs) have been reported to read, edit and write the weights of small neural networks, raising the prospect of auditing clinical prediction models f…