I’m Darko Stojilović
@ University College London
Here you can find a few examples of the research projects I have been involved in.
A gamified experiment
My recent work was focused on shaping effective AI policies, particularly concerning liability and accountability when autonomous agents cause harm to humans. I led the development of the gamified psychological experiment The Blame Game to study how different factors shape blame judgments to AI. The game used a visual novel format, combining text and images, allowing players to make choices along the way and to ultimately decide how blameworthy the suspect is (artificial agent or human agent).
In this project, I combined my expertise in methodology and experimental design with UX design principles. I presented the study results at several scientific conferences and published an article. The results of the study provided new insight for establishing better legal guidelines for assigning blame and liability in cases where AI causes harm.
Changing behaviour of young people
Confronted with the pervasive issue of low engagement among young people in democratic processes, I designed a comprehensive three-step project. Firstly, I conducted focus groups comprising diverse segments of young people. This enabled me to identify key underlying factors contributing to their disengagement. Secondly, in collaboration with the creative team, I developed three distinct campaign ideas based on the insights from the previous study. Finally, to determine the most effective campaign approach, I conducted a randomised controlled behavioural experiment with a new sample of young people. They were presented with the three campaign ideas and had to evaluate them across various dimensions, providing quantitative measures for each campaign.
The campaign gained considerable visibility through extensive media coverage, after which we conducted a new nationwide survey. Through the implemented campaign, we significantly increased the participation of young people in democratic processes. This achievement was evidenced by the changes observed in attitudes and behaviours, as evaluated by subsequent studies.
Data-driven insights to inform policy
I developed cutting-edge methodology that enabled the creation of quick and affordable surveys to obtain more relevant and precise insights into citizens' attitudes and behaviour. Instead of traditional lengthy questionnaires, I employed shorter, more focused surveys tailored to specific policy areas. To overcome contextual issues, I introduced a novel method of sampling, merging random digit dialling with pre-selection from a pool of all possible numbers nationwide to ensure a random sample.
By consistently surveying a nationally representative sample using these methods, we gained deep insights into societal attitudes and behaviours—ultimately providing meaningful recommendations to public policy decision-makers. For example, we were able to pinpoint a shift in public opinion regarding major political powers following significant changes in media reporting within a span of a couple of months.
Data analytics platform
I collaborated with a team of software engineers and designers to develop a comprehensive platform for real-time data analytics on Election Day. This project entailed programming automated statistical formulas to ensure efficient and accurate calculations, as well as crafting a user-friendly design based on feedback from field researchers. During the election day, the platform was used by more than a 1000 observers to input data, providing researchers with automated statistics and visual representations of data. This enabled the organisation monitoring the election process to be the first to report results on a national scale that day—outpacing even the national election agency.