BACKGROUND
There is widespread misinformation about the effects of alcohol consumption on health, due to a complex number of strategies used by the alcohol industry due to marketing, the lack of marketing regulation, the lack of political commitment from governments to tackle alcohol related problems, among other factors. This misinformation was amplified during the COVID-19 pandemic, with disastrous consequences early on, fueled by alcohol marketing activities online and in social media, following the massive use of various platforms by the public. Within this context, chatbots and conversational agents became an important piece of the WHO and PAHO response to the COVID-19 pandemic, developed to quickly disseminate evidence-based information.
OBJECTIVE
To develop and deploy a conversational agent to talk to users about alcohol related topics and recommend ways to reduce their risks from drinking.
METHODS
The content development was based on topics that are commonly covered by WHO and PAHO on alcohol related harms, misperceptions about alcohol related risks, social norms about drinking, and followed the latest scientific evidence on alcohol and health. The avatar was developed through a non-exclusive license agreement with a private company and included Google DigitalFlow ES as the natural language processing software, and AWS for cloud services. Another company was contracted to program all the conversations, following the technical advice of PAHO.
RESULTS
No identifiable data was used and therefore no ethical review was necessary according to PAHO's ethical review board. The conversational agent was named Pahola, speaks in English, Spanish and Portuguese, interacts anonymously to a potential infinite number of users through various digital devices. Pahola attracted good attention from the media, and potentially reached 1.6 million people, leading to 236,000 sessions on its landing page, mostly through mobile devices. The average time people effectively spent talking to Pahola was five minutes. Major dropouts were observed in different steps of the conversation flow. There were questions which she was not properly programmed to answer and some that no user ended up very interested in learning about.
CONCLUSIONS
Our findings demonstrate that Pahola could serve as a health educator on topics related to alcohol, not only in the Americas but globally. Improvements are needed to make the content completer and more robust. The potential of chatbots to educate the public on alcohol related topics but also other health risks and conditions seems enormous and needs to continue to be explored and expanded. At the same time, controlling alcohol marketing and preventing the alcohol industry from disseminating misinformation, challenging scientific facts and confusing the public will be fundamental to gather the most benefit from conversational agents.
CLINICALTRIAL
n/a