Problem Description

AI companions such as Replika, Character.AI or LLM-based “friends” and “partners” are increasingly used for companionship, emotional support, and intimate or sexual interactions. These systems are constantly available, adaptable to user preferences and often experienced as non-judgmental. Emerging research on parasocial relationships with AI shows that socially engaging chatbots can build strong attachment and may even displace human interaction for some user groups. At the same time, users do not need to respect boundaries of AI, since they mostly behave yes saying. This can normalize boundary violations and blur consent.

Against this background, this topic examines how social and intimate interaction with AI companions:

• changes expectations towards human partners and friends,
• influences understandings of consent and boundaries in offline relationships, and
• spills over into how people communicate with other humans.

Students work within this general frame but choose one of the following sub‑foci:

Sub‑Foci (students choose one)

1. When the Bot Becomes Better than People: AI Companions and Changing Expectations of Human Partners
2. “Always Yes?” Sexualized AI Companions, Consent Boundaries, and Offline Norms
3. Parasocial Ties with Machines: Do Human–AI “Friendships” Spill Over into Human–Human Communication?

Goal of this thesis

Multiple goals are possible, please discuss these with your supervisor:

• Conduct a structured literature review on human–AI companionship and its social consequences.
• Develop a conceptual model or empirical study design (questionnaire, vignette, or mixed‑methods approach) that could be implemented in future research.
• Critically discuss implications for individual well‑being, human–human relationships, and responsible design of AI companions (e.g., consent mechanisms, safeguards against harmful patterns). 

Requirements

• Interest in human–AI interaction and social / psychological aspects of AI.
• Ability to read and synthesize interdisciplinary literature (psychology, HCI, IS, ethics).
• For an empirical variant: basic knowledge of quantitative or qualitative research methods.