Papers by Dongjun Kang
From Values to Opinions: Predicting Human Behaviors and Stances Using Value-Injected Large Language Models (2023.emnlp-main)
Copied to clipboard
| Challenge: | Existing large-scale surveys soliciting opinions on issues can be costly and laborious. |
| Approach: | They propose to use value-injected large language models to inject a target value distribution into large language model (LLM) and have them predict opinions and behaviors of people with similar values. |
| Outcome: | The proposed method significantly outperforms baseline methods on four tasks and the results suggest opinions and behaviors can be better predicted using value-injected LLMs. |