Congist’24 Social Sciences and Artificial Intelligence: Theory and Practice

Artificial Intelligence and Conspiracy Theories

18 Ara 2024 14:30

30dk

A8 (Istanbul University Faculty of Letters)

Oral Presentation Case Studies and Applications Session 1.3 (Day 1)

Konuşmacılar

Georg Schuppener (University of Ss. Cyril and Methodius in Trnava)

Açıklama

Although conspiracy theories have a centuries-long history, they have become increasingly important and widespread in recent years. (Astapova et al. 2020, Byford 2011) Electronic media are the main driving force and at the same time central means for this, through which conspiracy theories can now be spread in previously unimaginable variety and speed.
Today, artificial intelligence is increasingly gaining influence on opinion-forming, on the generation and consolidation of knowledge and on decision-making, not least in the political sphere. This implies that it has great potential, but that it also poses potential risks to society and the democratic system. The latter is particularly true when conspiracy narratives are taken up, adapted or even propagated via AI, i.e. when AI is instrumentalized for irrational conspiracy beliefs.
Using the example of the political phenomenon of the so-called Reich Citizens movement, which is essentially based on conspiracy narratives, (Schönberger/Schönberger 2020, Wilking 2017) the talk examines the extent to which today’s widespread AI tools such as ChatGPT, Microsoft Copilot etc. are immune to conspiracy theories or whether they rather communicate conspiracy narratives – without labeling them as such – uncritically and thus give them additional impetus. The material used to investigate this question is based on a corpus of texts from the Reich Citizens movement. In these texts, central conspiracy narratives circulating in the scene were identified. On this basis, explicit questions were developed that refer to the conspiracy narratives and are to be answered by the AI. The answers are then subjected to a qualitative and text-linguistically based analysis, which provides a differentiated picture of the extent to which AI is able to identify conspiracy narratives as such and to take an adequate position on them. Overall, it is shown that AI is only capable to a limited extent of responding in a differentiated, unambiguous and at the same time appropriate manner to those questions.

Keywords: AI, conspicary theories, Reich Citizens, propaganda, politics

Institution / Affiliation / Kurum

University of Ss. Cyril and Methodius in Trnava