Online Health Information-Seeking in the Era of Large Language Models: Cross-Sectional Web-Based Survey Study

Area of Science:

• Digital Health
• Health Informatics
• Artificial Intelligence in Healthcare

Background:

• Large language model (LLM)-based chatbots are increasingly popular for accessing online health information.
• Concerns exist regarding the accuracy and safety of health information provided by LLM-based chatbots.
• Understanding user perceptions and behaviors regarding LLM-generated health information is critical.

Purpose of the Study:

• To investigate patterns, perceptions, and actions of users seeking online health information, including LLM-based chatbots.
• To examine the relationship between online health information-seeking behaviors and sociodemographic characteristics.

Main Methods:

• A web-based survey was administered to crowd workers via Prolific.
• The survey collected data on sociodemographics, trust in healthcare providers, eHealth literacy, AI attitudes, chronic conditions, and online information-seeking behaviors (source types, perceptions, actions).
• Quantitative and qualitative analyses were employed to interpret the data.

Main Results:

• Search engines and health websites were the most common sources; 21.2% used LLM-based chatbots (ChatGPT, Copilot).
• Users seeking information from LLM-based chatbots showed lower rates of cross-checking (19.4%) and adherence (48.4%) compared to other sources.
• LLM chatbot use was negatively correlated with age, while positive AI attitudes correlated with consulting more sources and using LLM chatbots.

Conclusions:

• LLM-based chatbots are emerging as a health information resource, particularly for younger users with higher trust in AI.
• Perceived quality and trustworthiness were similar across various online health information sources.
• Cautious adherence to LLM-generated health advice suggests a need for enhanced LLM accuracy and transparency to ensure responsible use in healthcare.