JMIR Publications

ABSTRACT

Background:

Large Language Model (LLM) chatbots such as ChatGPT may be able to provide support to people who smoke cigarettes and are trying to quit yet few studies have examined LLM chatbots.

Objective:

This pilot study examined the feasibility and acceptability of offering a specialized ChatGPT-based chatbot, BeFreeBot, integrated into a smoking cessation text-messaging program, BeFree.

Methods:

Participants who smoked cigarettes in the past 7 days were recruited from Amazon Mechanical Turk (N=23), enrolled in BeFree and provided access to BeFreeBot. Surveys were administered at baseline and again 4 weeks post-enrollment to assess perceptions of BeFreeBot. Computer records of interactions with BeFreeBot were also analyzed to assess participant engagement and BeFreeBot adherence to instructions.

Results:

Most participants (69.5%) texted BeFreeBot with questions or concerns at least once. Participants texted 14.5 (SD=23.6) texts to BeFreeBot on average. The majority of participants were highly satisfied with the BeFreeBot (72.2%) and agreed that it was helpful for quitting (57.9%). They also reported that the BeFreeBot responses were clear and easy to understand (94.1%), and that they trusted responses from the BeFreeBot (70.6%). The majority of participants (89.5%) reported trying to quit smoking for one day or longer, and 30.4% self-reported no smoking in the past 7 days. An analysis of transcripts of BeFreeBot responses (N=328) revealed that the BeFreeBot functioned as instructed on most measures, including on adherence to the US Preventive Services Task Force guidelines.

Conclusions:

A specialized LLM chatbot integrated into a text messaging program and accessed through text message was found to be feasible and acceptable to smokers.