| Popis: |
As conversational agents become more human-like, people expect them to be engaging as well. However, developing agents that comprehend human desires and generate appropriate responses, continues to be a challenge. We, therefore, conducted an interdisciplinary study and collected 2,300 human open-domain dialogs with self-labeled psychological variables such as empathy, connectedness, respect, and friendliness. We found that participants who talk coherently and disclose self-relevant information in the first encounter were regarded as engaging partners. Also, we found that various empathetic responses were critical for sincere interaction: agreement, perspective-taking, referring to someone as adorable and asking questions. When comparing the most and least engaging dialogs, linguistic cues and length of sentences denoted different extents of perceived empathy and sincerity by the partner. Additionally, we discovered that while large language models can generate impressive small talk in one-shot, they are limited in their ability to generate a variety of empathetic expressions and maintain a longer conversation. We propose a new approach for enhancing conversational agents' social and engaging characteristics. |
