Keywords: human values, ai safety, alignment, social impact, human-AI interaction
TL;DR: We present a new learning paradigm called Second Thoughts, that enables LMs to re-align with human values.
Abstract: We present Second Thoughts, a new learning paradigm that enables language models (LMs) to re-align with human values. By modeling the chain-of-edits between value-unaligned and value-aligned text, with LM fine-tuning and additional refinement through reinforcement learning, Second Thoughts not only achieves superior performance in three value alignment benchmark datasets but also shows strong human-value transfer learning ability in few-shot scenarios. The generated editing steps also offer better interpretability and ease for interactive error correction. Extensive human evaluations further confirm its effectiveness.
Supplementary Material: pdf