Explainable Persuasion for Persuasive Interfaces: The Case of Online Gambling

Conference: Bournemouth University, Faculty of Science and Technology

Abstract:

As human attention is a scarce resource, interactive online platforms such as social networks, gaming and online gambling platforms utilise persuasive interfaces to maximise user engagement. However, ethical concerns may arise since persuasive systems influence user behaviours. While interacting with persuasive systems, users may be unaware of being persuaded or unaware of the negative consequences that may result from interacting with persuasive systems. This can hinder users’ ability to evaluate the persuasion attempt and regulate their behaviour. Moreover, persuasive systems designed to maximise user engagement may, in some cases, trigger or reinforce addictive usage. There is evidence in the literature that online persuasive interfaces may influence psychological and cognitive mechanisms related to addictive behaviour. Transparency and user voluntariness are proposed to be the building blocks of ethical persuasive systems. However, to date, the concept of transparent persuasive technology mainly remained philosophical in academia. One approach to designing persuasive systems that adhere to the transparency and user voluntariness requirements could be fulfilling conditions for informed consent. When interacting with persuasive systems, users could be informed about the persuasive design techniques used by the system, and such information may help users build resilience against persuasion attempts made by the system. Such an approach aligns with the principles outlined in the software engineering code of ethics of avoiding harm and maintaining honesty and trustworthiness. This thesis aims to introduce and evaluate the concept of explainable persuasion in the context of designing ethical digital persuasive interfaces with an analogy to explainable artificial intelligence. A mixed methods approach was conducted to achieve this goal. The thesis focused on a distinct domain, online gambling, as gambling disorder is recognised as a mental disorder by health organisations. Accordingly, a scoping review was conducted first to identify the main persuasive design techniques utilised in online gambling platforms. Identified persuasive design techniques were analysed for their potential to facilitate gambling disorder through the addiction literature. An online survey was then conducted to examine users’ awareness of persuasive design techniques used in online gambling platforms and users’ attitudes towards the concept of explainable persuasion. Finally, an online experiment was conducted to determine the effectiveness of explainable persuasion as an inoculation intervention in building resilience against persuasive design techniques used in online gambling platforms. The findings of the user studies showed that explainable persuasion was accepted and that it could be a promising solution for designing persuasive interfaces that promote informed choice and strengthen resilience against persuasion if it is not compatible with users’ personal goals. This thesis contributes to transparency and explainability literature as it is one of the first attempts to examine the role of explainability in the domain of persuasive technology which may also have addictive potential. Identifying acceptance and rejection factors of explainable persuasion can help design persuasive interfaces that promote informed usage and meet ethical requirements. This implication does not only apply to persuasive technology but can also be generalised to research areas such as combatting fake news and social engineering. The findings are expected to have important implications for gambling operators and regulators in expanding the scope of responsible gambling practices to ensure explainability and transparency. The results are expected to also benefit wider application areas such as explainability in other contents and interfaces related to marketing, news and recommendations made by or facilitated by intelligent systems.

https://eprints.bournemouth.ac.uk/38901/

Source: Manual