AI for addictive behavioural intervention to archive population effects

ABSTRACT

Background: Smoking, harmful alcohol use, and mental health problems affect many people and create major challenges for public health. Many people want to change these behaviours, but it can be difficult to get timely, personal, and continuous support. Digital health tools, such as apps and chatbots, can reach large numbers of people, but many users stop using them quickly or do not find them personal enough. Generative artificial intelligence (GenAI), such as AI systems that can produce human-like conversations, may offer a new way to support people who want to quit smoking, reduce alcohol use, or improve their mental wellbeing. GenAI could provide instant responses, personalised encouragement, practical advice, and follow-up support at scale. However, we still need to understand how to design these tools well, how to keep people engaged, and how to use them safely and effectively in real life.

Purpose: This research aims to explore how GenAI can be used to develop better interventions for addictive behaviours, especially smoking cessation and alcohol reduction, while also supporting mental health. Based our many trials in these areas, the program aims to examine how AI can be used not only as a standalone chatbot, but also as part of a hybrid support model. GenAI may work together with human counsellors, health professionals, and community services. The long-term goal is to find ways to make behaviour change support more accessible, engaging, and effective for large populations.

Research questions:
(1) How should GenAI interventions be designed so that they give accurate, helpful, and supportive advice?
(2) What kinds of messages, functions, and interaction styles can encourage people to keep using the intervention?
(3) How can AI conversations be combined with support from human counsellors and existing health or social services?
(4) How can these interventions help people change their behaviour, for example by increasing motivation, confidence, coping skills, and emotional support?
(5) What types of studies are needed to test whether GenAI interventions are effective, safe, fair, affordable, and scalable?
By answering these questions, this research can help us understand how AI, behavioural science, and public health can work together to address addictive behaviours and improve population health.

SKILLS & EXPERIENCE REQUIRED

Background knoweldge on GenAI and human behaviors, basic statistics, and literature review