Chatbots are non-human agents, programmed using artificial intelligence (AI) to talk to a person as if a real-life conversation is taking place. It’s common to visit a website and be greeted with an invitation to chat about its commodities – whether suggesting products, offering technical support, or assessing customer satisfaction, chatbots have established their place in assuming a role traditionally undertaken by humans.
Healthcare is no exception. Here chatbots have been used to educate, diagnose, and support a technically savvy and demographically diverse population. The use of chatbots in addiction services has yet to be appreciated however, with research showing little traction in this area in the last five years. This is contrary to the upward trend seen in other areas of healthcare, for example with anxiety, depression, and obesity. This gap is perhaps most surprising given the target population and model of interaction offered – interaction that’s free of judgement and prejudice and that can reach people reticent to openly discuss their problems. This gap presents an untapped opportunity, a blank canvas for developing chatbots to serve those needing support with drug and alcohol addiction, including those in recovery.
Point of need
It is here we introduce Foxbot, a recovery friend. Foxbot was born from the idea to personify the online delivery of recovery interventions – interventions that combine recovery knowledge with positive psychology, and concepts from positive computing, to use technology to improve wellbeing. The basic design requirements for Foxbot were to be friendly and easy to form a connection with, to allow users to direct their own experience based on what they’re facing and how they’re feeling, and to provide a fresh, accessible, and engaging way of supporting people in recovery.
Foxbot’s purpose is to help people at their point of need by supporting them with some of the common difficulties encountered in recovery. Foxbot achieves this by offering suggestions, and quick and easy interventions, that can be used and re-used to bolster recovery. Unlike more conventional chatbots that form part of a larger online strategy to promote or sell products and services, Foxbot’s intentions are entirely altruistic. Foxbot’s goal is to connect with and understand what people in recovery are commonly challenged with and respond with an intervention that can help them in the present moment.
Feeling supported through connection and understanding is highly valued in recovery, so these characteristics were considered rudimentary to Foxbot’s design, and embedded through the content of his knowledge base and the relaxed style of communication he’s programmed to use. It was decided that if Foxbot was to be a genuine and successful recovery friend, he should exhibit his own character strengths – social intelligence, self-regulation, gratitude, humour, and creativity. Social intelligence for fluent social function and mature judgement; self-regulation for limitless patience and satisfaction with friendly relationships; gratitude to express positive emotions and optimistic prosocial behaviour; humour to promote positive mood and endurance, and creativity for his novel delivery of recovery interventions. These signature strengths were programmed into Foxbot’s communication patterns to enable him to support people who wish to talk about things like having cravings, being in a bad mood, needing a reality check on what it means to be in addiction, having been triggered, or looking for a recovery boost.
The interventions delivered by Foxbot take common experiences from recovery and amalgamate them with positive psychology theory to offset and enhance the user’s current frame of mind. For example, if someone tells Foxbot they’re having a craving, he does two things – firstly, he suggests a mindful practise for letting the craving pass, one that uses positive imagery to represent the craving as a process that has an end. Secondly, he involves the user in an activity that will refocus them. In this case, an online matching game, where they do not need to feel self-conscious about participating, one that gives them feedback, and is rewarding. This intervention equips the user with a strategy to deal with cravings, where they are encouraged to participate in an activity that distracts them after visualising the craving as a process that comes to an end.
Another common experience for people in recovery is having unhelpful thoughts – that they don’t really have a problem with alcohol or drugs, and that everything will be okay this time. Here Foxbot helps by offering a reality check in the form of an adage that can challenge the unhelpful inner voice with a relatable example of something that’s apparent with the hindsight of recovery. For example, ‘you can’t drink away alcoholism’ or ‘do you recall sobering up when you patted your pockets and couldn’t find your phone’.
If you tell Foxbot you are in a bad mood, he will attempt to raise a positive emotion using humour. Whilst his sense of humour may have been influenced by his design team, the jokes have been tested on a wider audience before they made the cut.
If a user is not experiencing a particular challenge but would still like to do something positive for their recovery, Foxbot offers the option of receiving a recovery boost. Research has shown that life appreciation and gratitude play a fundamental role in sustained and successful recovery. Foxbot capitalises on this by offering recovery boosters, powerful reminders of the things that are easy to take for granted or lose sight of in recovery, for example, ‘see you later hangover’, ‘welcome back energy’, ‘adios shaking hands’ and ‘goodbye bloodshot eyes’. This is to remind users of the positive outcomes they’ve acquired thanks to living in recovery, outcomes that can only be preserved by sustaining that recovery.
By providing an unlimited willingness to engage in conversation, Foxbot frequently relays the message that talking is important to recovery. His conversational style includes phrases such as ‘It’s good to talk’, ‘talking to you brightens my day’, and ‘don’t struggle in silence’. Foxbot also gauges how useful an intervention is proving to be through the acquisition of real-time feedback, where, if necessary, an extended version of the intervention is presented to the user. For example, if a user has asked for a recovery boost, Foxbot will check-in during the intervention to ask if they feel they have had a boost, offering feedback options of ‘yes, boosted’ or ‘no, power me up’. If ‘no’ is selected, another set of recovery boosters are presented to the user.
To keep the conversation fresh, and avoid anticipated chat, Foxbot has a continually developing knowledge base. The AI component of his programming assigns different conversational experiences to the user to avoid ‘hackneyed dialogue.’ As an example, when a user opts to ‘chat again’, a randomised response is given, such as, ‘talking opens your mind to new ideas’ or ‘it’s always a good time to talk to me about recovery’. This ensures the user experience is unpredictably different with each encounter.
Foxbot can be considered as being in his formative years – he is undergoing significant developmental change, which includes a growing knowledge base and repertoire of recovery suggestions. In addition, Foxbot can also remember his past conversations, so as he matures historic chats will be analysed to better inform how he should evolve. For example, if a disproportionate number of users engage in a specific intervention, this area will be prioritised with greater investment made to enrich the support available. Foxbot will also become wiser by gathering feedback on his own strengths, including his performance as a recovery friend. When he reaches this level of maturity, he – like his recovery friends – will be learning how to better use and develop his own strengths.
To get to know Foxbot and find out what else he can do, visit positivelysober.org where you can also look at the work being done to create a positive approach to recovery.
Lisa Ogilvie is a PhD student, Julie Prescott is reader in psychology, and Jerome Carson is professor of psychology, all at the University of Bolton