Pierre PetroninFollow


The data quality of open-ended answers gathered via online surveys can be low because of a large proportion of blank, short, or ambiguous responses. However, online surveys do not provide an opportunity to resolve ambiguities or seek elaboration by prompting respondents for additional information. While conversational surveys can overcome this limitation, they are generally not suitable for close-ended questions. This disclosure describes techniques to augment open-ended questions within traditional online surveys by embedding conversational capabilities enabled via an AI agent or chatbot that can be based on a large language model. The conversational AI agent can automatically generate follow-up questions based on responses to open-ended questions. The techniques utilize an appropriately trained conversational AI agent (or chatbot) to automatically include an open-ended question type, e.g., using an embedded data field of type “conversation,” along with relevant variables and scripts to display the conversational experience within the survey webpage, app, or other interface, and to enable information exchange between the conversational interaction and the online survey.

Creative Commons License

Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.