Hiding in Plain Site: A Turing Test on Fake Persona Spotting

Mckenzie, Grace and Taylor, Paul and Conchie, Stacey (2024) Hiding in Plain Site: A Turing Test on Fake Persona Spotting. PhD thesis, Lancaster University.

Text (2024McKenziePhD)
2024McKenziePhD.pdf - Published Version
Download (19MB)

Abstract

This research investigated the ability of humans to accurately detect fake Facebook profiles. The prevalence of fake profiles on social media provides a consistent threat to users from malicious actors and computer-generated identities. This study wanted to move away from software and algorithm focused attempts to counter these threats and instead put the emphasis on the individual and their ability to detect a fake profile. Participants were shown a series of fake Facebook profiles (created specifically for this research) and real Facebook profiles and tasked with judging the authenticity of said profiles. Participants were also asked to identify the areas of the profile they had used to make their decision using heatmap software. Across the six studies within this research, new experimental manipulations were introduced each time in the form of time pressure, cross-cultural profiles, and training interventions. This approach allowed for investigation into the conditions that have the greatest influence over participants judgement accuracy, and also aligned the studies more closely with real world aspects of decision making in the online world. Participants were more accurate at correctly identifying real profiles as real than they were at correctly identifying fake profiles as fake, and fake profiles with a higher number of manipulated characteristics (4 Fakes) were judged more accurately than those with fewer (2 Fakes) or zero characteristics (0 Fakes). However, their judgment accuracy was often no better than that of chance. Additionally, participants in all studies relied heavily upon the visual stimuli of the profiles (manipulated characteristic Photo Type) to inform their authenticity judgements. The insights gathered from this study add to the literature around online deception, establish a foundation for future research with a person-centric approach, and inform the design of future studies exploring similar themes.