Training Socially Aligned Language Models on Simulated Social Interactions
Abstract
The goal of social alignment for AI systems is to make sure these models can conduct themselves appropriately following social values. Unlike humans who establish a consensus on value judgments through social interaction, current language models (LMs) are trained to rigidly recite the corpus in social isolation, which causes poor generalization in unfamiliar cases and the lack of robustness under adversarial attacks. In this work, we introduce a new training paradigm that enables LMs to learn from simulated social interactions. Compared with existing methods, our method is much more scalable and efficient, and shows superior performance in alignment benchmarks and human evaluation.
Cite
Text
Liu et al. "Training Socially Aligned Language Models on Simulated Social Interactions." International Conference on Learning Representations, 2024.Markdown
[Liu et al. "Training Socially Aligned Language Models on Simulated Social Interactions." International Conference on Learning Representations, 2024.](https://mlanthology.org/iclr/2024/liu2024iclr-training/)BibTeX
@inproceedings{liu2024iclr-training,
title = {{Training Socially Aligned Language Models on Simulated Social Interactions}},
author = {Liu, Ruibo and Yang, Ruixin and Jia, Chenyan and Zhang, Ge and Yang, Diyi and Vosoughi, Soroush},
booktitle = {International Conference on Learning Representations},
year = {2024},
url = {https://mlanthology.org/iclr/2024/liu2024iclr-training/}
}