Abschlussarbeiten

In the evolving landscape of social media, the emergence of virtual and AI-driven influencers marks a significant paradigm shift, challenging traditional notions of interaction and influence. This research aims to delve into three critical aspects of this phenomenon: the nature of interactions with AI influencers, the implications of these entities on users' social media well-being, and the dynamics of para-social relationships formed with virtual/AI entities.

Topics may include, but are not limited to, the following

• Social Media Intergroup Contact with Virtual/AI Influencers: The investigation centers on the application of intergroup contact theory to the domain of social media, particularly in interactions involving AI influencers. It seeks to understand the role of virtual or AI entities, designed to mimic human beings in social media, in influencing intergroup relations. The primary objective is to evaluate the potential of AI influencers in reducing prejudice under the framework of intergroup contact theory.
• Well-being in Social Media with Virtual/AI Influencers: The use of social media can have both positive and negative effects on the well-being of its users. In recent years, the emergence of virtual/AI influencers, who are increasingly indistinguishable from real people on social media platforms, has introduced new dynamics in digital interactions. This study aims to explore the effects of these virtual/AI influencers on the well-being of social media users, examining how their presence may impact mental health, self-perception, and overall well-being in and beyond the digital sphere.
• Para-social Relationships with Virtual/AI Entities: AI entities such as Snapchat's My AI, ReplikaAI and Celebrity AI are designed to form intimate bonds with real people, often targeting areas such as both romantic and friendship bonds while monetizing digital loneliness. The study of this phenomenon addresses the nature of the parasocial relationships between users and these AI agents and aims to understand the psychological dimensions, the impact on users and the wider implications for future human-computer interactions.

These studies represent a comprehensive effort to understand the multifaceted effects of virtual and artificial agents through social media. By exploring these diverse areas, the research aims to provide valuable insights into the complex interplay between virtual agents, para-relationships, and individual well-being.

Requirements & Contact:

Students should have an interest in social media analytics, sociological theory, and mixed-methods research.

Further information, detailed references, and specific keywords related to each area of study can be provided upon request, ensuring a comprehensive and tailored understanding of these complex social media phenomena.

If you are interested on writing a thesis on Social Cohesion and Digital Democracy, please reach out to Georg Voronin.

Description:

Artificial Intelligence (AI) is being used more and more these days by companies and individuals for a wide variety of tasks which can be best seen through the popularity of ChatGPT and similar (AI)-based technologies. One of the areas in which AI is being applied is to recognize patterns in different types of data, which can span from formats such as images, and videos to textual data. Further, these technologies find application in different contexts such as in medicine to recognize diseases in medical images (Chan et al., 2020) or text processing in the hiring process with AI-based CV screening software (Albert, 2019). However, one frequent criticism of the use of AI is the lack of explanation of its results due to the black box principle that often surrounds the AI technologies. This black box principle describes a situation where it is not possible to see the inner workings of a technology meaning that it is not feasible to follow how a given input to a technology results in the technology’s output which tests the individual's trust in the technology and the correctness of its results. This in turn led to the development and spread of explainable AI (XAI) techniques such as SHapley Additive exPlanation (SHAP) or Local Interpretable Model-Agnostic Explanations (LIME), which can be used to attempt to explain the results of different AI approaches.

In the field of cyber security, communication online for example on social media platforms such as Twitter about cyber threats is a valuable resource for identifying relevant incidents (Eyilmez et al., 2020). Moreover, AI-based approaches for the detection of cyber security-related events online are proving to be successful and promising approaches for the purpose of cyber threat identification (Sceller et al., 2017).  In the context of XAI’s effect on the acceptance of AI results research results are inconclusive due to different research findings indicating an increase in acceptance while others don’t (Schemmer et al., 2022). This begs the question of whether XAI explanations have an effect on the acceptance of AI-detected cybersecurity-related events. Due to the public not being very familiar and well-versed in the topic of cyber security while it is becoming an increasingly important topic to society the effect of XAI on the acceptance of AI-identified events compared to human-identified incidents could differ from the acceptance of AI-identified events in different domains. This research could help in increasing the public's willingness to follow warnings of cyber threats through adjustments based on the findings of this research.

Therefore, in this thesis, different messages will be designed using AI and human-identified incidents. Furthermore, a distinction between AI-based identification with an XAI component and without an XAI component shall be made. These are to be visualized and evaluated in the form of mockups or a click dummy. The evaluation will be based on an online study and incorporate recent knowledge from literature (e.g., Eyilmez et al., 2022; Sceller et al., 2017; Schemmer et al., 2022; Riebe et al., 2023, Basyurt et al., 2022).

Requirements & Contact:

For this thesis you should be interested in cyber security, Artificial Intelligence and ideally have worked with quantitative research data before.

If you would like to apply for this thesis, please contact Ali Sercan Basyurt

References:

Albert, E. T. (2019). AI in talent acquisition: a review of AI-applications used in recruitment and selection. Strategic HR Review, 18(5), 215-221.

Basyurt, A. S., Fromm, J., Kuehn, P., Kaufhold, M. A., & Mirbabaie, M. (2022). Help Wanted-Challenges in Data Collection, Analysis and Communication of Cyber Threats in Security Operation Centers.

Chan, H. P., Samala, R. K., Hadjiiski, L. M., & Zhou, C. (2020). Deep learning in medical image analysis. Deep Learning in Medical Image Analysis: Challenges and Applications, 3-21.

Eyilmez, K., Basyurt, A., Stieglitz, S., Fuchss, C., Reuter, C., & Mirbabaie, M. (2022). A Design Science Artefact for Cyber Threat Detection and Actor Specific Communication.

Riebe, T., Biselli, T., Kaufhold, M. A., & Reuter, C. (2023). Privacy Concerns and Acceptance Factors of OSINT for Cybersecurity: A Representative Survey. Proceedings on Privacy Enhancing Technologies, (1), 477-493.

Sceller, Q. Le, Karbab, E. M. B., Debbabi, M., and Iqbal, F. 2017. “SONAR: Automatic Detection of Cyber Security Events over the Twitter Stream,” ACM International Conference Proceeding Series.

Schemmer, M., Hemmer, P., Nitsche, M., Kühl, N., & Vössing, M. (2022, July). A meta-analysis of the utility of explainable artificial intelligence in human-AI decision-making. In Proceedings of the 2022 AAAI/ACM Conference on AI, Ethics, and Society (pp. 617-626).

Description

The emergence of mobile devices such as smartphones and wearables has transformed the healthcare industry, enabling patients to self-manage their health. Mobile app platforms such as the Google Play Store and the Apple App Store have become key players in the mobile health (mHealth) domain (Gleiss et al., 2021). Those apps collect much data, that might one the one hand improve their outcomes, but might also induce privacy risks that are especially crucial in the health domain. From previous research it is known, that there is a phenomenon called the “privacy paradox” (Kokolakis 2017), which posits that even though users say that they value their privacy, they do not act accordingly. Some researchers propose a privacy calculus perspective (Wang et al. 2016) on this issue, which implies that users weigh the perceived benefits and risks before making a decision.

One other explanation could be, that collecting data and sharing it with others might lead to the app becoming part of a larger ecosystem of developers, third-party functionalities and other applications. This would take into account generativity (Fürstenau et al. 2023), i.e., that developers that are not part of the organization can be part of the wider ecosystem of an app (social view on generativity), as well as it is possible to integrate and develop new products and features (product view on generativity) by using and sharing data.

To do so, pre-existing data on data collection from the Apple App Store, as well as data from Github can be used to identify privacy and generativity. The thesis will provide a first conceptual view on a potential privacy-generativity-tradeoff as well as preliminary empirical evidence.

Requirements & Contact

For this thesis you should be interested time-series analysis and have some previous experience with Python. You should be interested in digital platforms as well as ecosystems. If you would like to apply for this thesis, please contact Kai Schewina.

References

Anderson, C. L., & Agarwal, R. (2011). The digitization of healthcare: boundary risks, emotion, and consumer willingness to disclose personal health information. Information Systems Research, 22(3), 469-490.

Fürstenau, D., Baiyere, A., Schewina, K., Schulte-Althoff, M., & Rothe, H. (2023). Extended generativity theory on digital platforms. Information Systems Research, 34(4), 1686-1710.

Kokolakis, S. (2017). Privacy attitudes and privacy behaviour: A review of current research on the privacy paradox phenomenon. Computers & security, 64, 122-134.

Wang, T., Duong, T. D., & Chen, C. C. (2016). Intention to disclose personal information via mobile applications: A privacy calculus perspective. International Journal of Information Management, 36(4), 531-542.

Description:

Large Language Models (LLMs) such as ChatGPT have the capability to generate scientific text, which is likely to be increasingly adopted in research. Current estimates suggest that at least 10 percent of scientific literature has been assisted by LLMs, with significant implications. This bachelor thesis should answer the question that field of science adopted LLMs the most, by analyzing the abstracts of journals different fields.

Requirements:

The following skills are required in order to succeed:

• Proficiency in at least one programming language, with a preference for Python.
• Knowladge in or willingness to learn fundamental and advanced techniques of natural language processing
• Interrest in the research domain.

Additionally you need to be enrolled in a Bachelor programm at the University of Potsdam.

Further information can be provided upon request.

If you are interested on writing a thesis in this field, please reach out to Till Schirrmeister

Description:

Knowledge creation is based on theories from many scientific fields, such as neuroscience, philosophy, computer science, information systems, and sociology. As artificial intelligence becomes more involved in scientific work, it starts to become an actor on its own, significantly affecting knowledge creation. The aim of this thesis is to summarize the existing theories in the field of knowledge creation, to develop a classification system from this analysis, and to explore the implications of AI entering the field of knowledge creation.

Requirements:

The following skills are required in order to succeed:

• Willingness to read a lot of challanging literature
• Endurance
• Interrest in the research domain

Additionally you need to be enrolled in a Master programm at the University of Potsdam.

Further information can be provided upon request.

If you are interested on writing a thesis in this field, please reach out to Till Schirrmeister