Workshop paper “The Reverse Turing Test: Being Human (Is) Enough in the Age of AI” published

On 7 June 2022, our colleague Fatemeh Alizadeh presented the workshop paper “The reverse Turing test: being human (is) enough in the age of AI” at the Sixth International Workshop on Cultures of Participation in the Digital Age, which this year raised the question “AI for Humans or Humans for AI?”. The conference paper was published in the Proceedings and can be found here.

Abstract:

Disposing of bad actors on social media is a daunting task, particularly in the face of “engineered social tampering” [4]. That is what Ferrara et al. [6] have labeled the rise of social bots, and large platform owners are struggling to mitigate the harmful effects caused by such malicious software. Therefore, it is no surprise that platform owners like META are fastening their security controls and that the popular press has tracked the efficacy of these measures. Specifically, META has been implementing what Forbes’ Lance Eliot named the ‘Upside Down Turing Test.’ [26]. Unlike the original Turing test, which tasked a human participant with distinguishing a human from a digital speech correspondent, this version is designed to use a software program to distinguish non-human activity on the platform. In this work, we discuss the complications introduced by this reversal taking the human user’s perspective. On the one hand, we recognize the necessity for fraud detection and defense against web-automated attacks. On the other hand, we find it necessary to uplift the voices of users who are wrongfully made victims as a result, in minor or major ways. At the same time, we offer alternatives to these invisible Reverse Turing Tests (RTTs) that expand the scope for distinguishing between human and non-human actors, while keeping humanity at the forefront of this inquiry.