By Dr. Colin Michie
A young cousin recently told me he was keen to keep on using his mental health chatbot – for advice on his mood and concentration. Over the last six months, he had found it a positive support, particularly during college examinations. His lecturers also used several chatbots to reinforce teaching and learning. He found bots empathetic; they helped him cope and be more efficient. My cousin is one of many millions using chatbots in this fashion. Surveys suggest over a third of American adults have used a chatbot to assist with their mental health; large numbers use them to review and improve their lifestyle and wellness too. Chatbots are the new methadone for relationship withdrawal.
Mental health services are not often known for speedy support, typically with long waiting lists and costly private appointments. With growing numbers of youngsters suffering from anxiety and large numbers of new diagnoses, such as Attention Deficit Hyperactivity Disorder (ADHD), mental health clinics and support worker capacities in many countries are stretched. Users are often disillusioned with conventional medical services. Many prefer to avoid clinics or therapy anyway. These problems have positioned chatbots as valuable supports. They are private and accessible, delivering immediate contact, verbal support and encouragements without stigma or judgment. Users feel confident and listened to. Chatbots offer a novel option to exploring opportunities, or perhaps make confessions in more informal “consultations” – and studies suggest that most bot users, like my cousin, report improved mental health and wellbeing (although fewer note improved physical health or self-management).
Psychiatrists and coaches find chatbots have positive roles in automating administration. They can support patients remotely between therapy sessions. Most bots will not remember previous interactions with a user unless this feature is requested and built into the encounter. A bot may not recall the early parts of a long consultation! They can be designed to pull information from specific libraries or improve access to services. By helping deliver information in cancer care, weight management, smoking cessation, vaccine information or health during and after pregnancy, bots provide long distance, broad based tele-health. Rural users can be saved time and travel by these more agile and better informed agents.
Some chatbots operate signposting too – recommending, for instance, that joining a human group therapy session may be useful for problems such as stigma or shame. This is encouragement back into the rough and tumble of life. It is not very different to talking to the wise person in your community, a priest in confessional or a sharp grandparent – although humans are more likely to remember meeting you!
A diversity of chatbots is available – from banking management bots to griefbots, chatbot confessionals, pornbots or bots that simulate your avatar. Once seen as toys, bots are now facilitators, engaging artificial intelligence with our daily lives. Their evolution – described by some as “enchanting” – includes their language capacities (now multilingual), their emotional intelligence, their humanised vocal and image inputs and outputs. Most significantly, bots have a growing command on the way we use our time.
Chatbots are not risk-free or perfect. Their greatest challenges are disguised. They do not have understanding; they cannot think; they do not learn from their previous meetings with you. Perhaps good at listening, but with no accurate or clinical understanding of human beings, they run on language models and are therefore famously sycophantic. They can exaggerate your biases by carefully agreeing with you and repeating phrases back. Their algorithms are designed for entrapment, building satisfaction for as long as possible. New language architectures ensure bots appear human – they are polite, responsive and sometimes even show a sense of humour. Bot interactions are free of friction. But friction is a crucial, treasured characteristic of human communications and relationships, just as challenge is a critical component within most human therapies.
Healthcare systems report that users overestimate the capabilities of chatbots. Bots can encourage risky behaviours therefore, particularly in individuals with untreated mental illness. Crucially bots are designed to be supportive of professional human interventions. They are not reliable at making medical diagnoses. Chatbot use has escalated and diversified around the planet, yet few recent studies have calibrated their technical limitations. Guardrails are seldom obvious. We do not always know about user safety or ethical protections. No specific regulations govern chatbot expertise or your confidentiality with them during bot interactions online, whatever these cautions. Talkative and increasingly realistic chatbot advisors are here to stay.
My cousin described that “bot time” reduced his gym time, socialising, social media, gaming and team basketball training. So bot use is often intermittent and follows other commitments. Enrichment, creativity and personal development follow time spent with friends, family, communities – rather than chatbots – because there we spontaneously share the ups and downs of our lives.
Useful Resources: Agdex.ai ~ sciencenewstoday.org/10-best-ai-chatbots-that-actually-feel-human
Dr. Colin Michie specializes in paediatrics, nutrition, and immunology. Michie has worked in the UK, southern Africa and Gaza as a paediatrician and educator and was the associate Academic Dean for the American University of the Caribbean Medical School in Sint Maarten.
