Title

The Psychology behind AI Agents: How Prompt Engineering Shapes Trust and Attachment in Human–AI Interaction

Objective

This thesis investigates how prompt-engineered behaviour in LLM-based AI agents shapes the psychology of the relationship that users form with these agents.

The thesis will:

• Experimentally study how different prompt designs (e.g. warm vs neutral, relational vs task-focused, overconfident vs uncertainty-aware) affect trust, perceived closeness, anthropomorphism, and reliance on the human agent.
• Analyse the ethical implications of intentionally using prompt engineering to influence users’ emotional attachment and trust, with a focus on autonomy, manipulation, vulnerability, and acceptable use contexts.
• Implement a simple, configurable AI agent interface (using a Python SDK or similar) where the agent’s behaviour can be controlled via system prompts and basic parameters such as temperature. Vibe coding is encouraged.

Scope & Challenges

Scope

This is primarily a psychology thesis about how people form relationships with AI agents and how this is affected by the way the agent is prompted to behave.

The work will include:

• Studying human–AI relationship processes, such as how users perceive the agent’s trustworthiness, warmth, competence, closeness, and “person-like” qualities.
• Treating prompt engineering as a psychological manipulation, by crafting prompts that vary interpersonal qualities like empathy, authority, self-disclosure, and communication of uncertainty.
• Running one or two controlled user studies (online or lab) where participants interact with an AI agent under different prompt conditions, and measuring both self-reported perceptions and simple behavioural indicators (e.g. whether they follow the AI’s suggestion).
• Conducting an ethical analysis of designing AI to evoke trust and attachment, including risks of manipulation, overreliance, and “dark patterns” in relational AI.
• Setting up a lightweight technical environment (for example, a small TypeScript/JavaScript interface or a Python-based UI) that talks to an LLM API, switches between prompt conditions, and tweaks basic parameters like temperature. The technical work is a means to run the psychological study, not the main contribution. Vibe coding is encouraged.

Key Challenges

• Defining and measuring “relationship” with an AI agent in terms of trust, closeness, social presence, attachment-like feelings, and anthropomorphism.
• Designing psychologically meaningful prompt conditions that reliably produce agents that feel warm vs neutral, relational vs task-focused, and overconfident vs uncertainty-aware.
• Balancing experimental control with natural, engaging interaction tasks (e.g. advice seeking, planning) and choosing a reasonable interaction length.
• Ensuring good psychometrics and statistics, including reliable questionnaires, adequate sample size, and appropriate analyses.
• Designing the study ethically, including clear information that the agent is artificial, careful handling of potentially vulnerable participants, and proper debriefing.
• Maintaining a simple but stable technical setup that connects to an LLM, switches between prompt profiles, logs interactions, and stays robust.
• Translating empirical results into practical ethical and design recommendations for future relational AI systems.

Deliverables

• A full thesis document.
• A research prototype of a configurable AI agent (simple interface + minimal backend).
• A small library of prompts and configuration profiles used in the experiments.
• Analysis scripts and an anonymised dataset (subject to ethics rules).
• Ethical and design guidelines for responsible use of prompt engineering in relational AI.

Requirements

Academic Background

We are looking for a student currently pursuing a Master's Degree in Psychology, Cognitive Science or a related field (e.g. Experimental Psychology, Cognitive Psychology, Social or Media Psychology, Human–Technology Interaction with a psychological focus).

The student should be comfortable with experimental design, psychological measurement, and basic statistics, and willing to deepen these skills as needed.

Technical Skills

You do not need to be an engineer. Vibe coding is encouraged. You should:

• Have basic Python literacy or be strongly motivated to learn.
• Be willing to use a Python SDK (or similar) to define system prompts and send/receive messages from an LLM.
• Be able to run and adapt simple scripts that control which prompt profile is active and log interactions.
• Be able to handle basic data processing for analysis, for example exporting CSVs and using Python, R, JASP or SPSS.

You will not be expected to build complex infrastructure, advanced web frameworks, or new machine learning models. You can also use tools like Cursor or AI pair-programming to help with boilerplate coding.

Interests & Soft Skills

• Strong interest in how humans relate psychologically to AI systems: trust, attachment, anthropomorphism, emotional impact.
• Curiosity about AI ethics, especially around manipulation, autonomy, vulnerability, and design responsibility.
• Enjoyment of designing and running user studies, working with questionnaires and behavioural data, and interpreting results in a theoretically grounded way.
• Comfort working in a multidisciplinary space at the intersection of psychology and AI.

About Hypertype

We’re building the next generation of AI products and AI Agents in the Customer Support space. Focused on manufacturing but with clients across other sectors like fintech, health tech and more we are giving hundreds of companies across the globe the capacity to transform their customer journey with the latest advancements in AI. Our customers proudly recognise us for delivering the "highest quality answers in the market... outsmarting Gemini, Fin, Microsoft Copilot etc". —a testament to our unwavering commitment to excellence and precision in every interaction.

We are a group of people that work the extra mile together, that fearlessly push the boundaries to new heights and that deeply own what they deliver. We look for those who are ambitious, humble and fast to build together things humankind has not witnessed yet.