Skip to main content

LUP Student Papers

LUND UNIVERSITY LIBRARIES

The Impact of Prompt Phrasing on AI-based Chatbot Effectiveness and User Experience

Brown, Kasper ; Elf, Arvid and Henriksson, Edvin LU (2025) SYSK16 20251
Department of Informatics
Abstract
This bachelor’s thesis explores how everyday users, specifically non-experts in higher education, interact with AI-based chatbots through natural language prompts. While prior research has focused on improving model performance and understanding prompt mechanics, little attention has been given to how users without technical expertise experience and shape these interactions. The study aimed to explore how different ways of phrasing prompts influence chatbot responses and user perceptions, ultimately generating actionable recommendations to enhance the prompting experience. Using a qualitative research design, data were collected from six university students through a workshop and interviews. Thematic analysis revealed six key findings... (More)
This bachelor’s thesis explores how everyday users, specifically non-experts in higher education, interact with AI-based chatbots through natural language prompts. While prior research has focused on improving model performance and understanding prompt mechanics, little attention has been given to how users without technical expertise experience and shape these interactions. The study aimed to explore how different ways of phrasing prompts influence chatbot responses and user perceptions, ultimately generating actionable recommendations to enhance the prompting experience. Using a qualitative research design, data were collected from six university students through a workshop and interviews. Thematic analysis revealed six key findings highlighting that prompt formulation is not a static act but a dynamic, iterative process. Users engaged in trial-and-error, adjusted tone and detail, assigned roles to the chatbot, and applied emotional or stylistic framing to guide responses. Prompting emerged as a form of active sensemaking and problem-solving, where users learned to refine inputs to achieve better outcomes. Drawing on Sensemaking theory, the study concludes that users are not passive recipients but active co-creators in AI interactions. The bachelor’s thesis provides practical recommendations such as encouraging iterative prompting, meta-prompting, emotional framing, goal alignment, and the use of templates to support more effective chatbot use. (Less)
Popular Abstract (Swedish)
Denna kandidatuppsats utforskar hur vardagliga användare, specifikt icke-experter inom högre utbildning, interagerar med AI-baserade chattbottar genom naturliga språkprompter. Tidigare forskning har främst fokuserat på att förbättra modellernas prestanda och förstå promptmekanismer, men har i liten utsträckning uppmärksammat hur användare utan teknisk expertis upplever och formar dessa interaktioner. Studien syftade till att undersöka hur olika sätt att formulera prompter påverkar chattbottens svar och användarnas upplevelse, med målet att generera handlingsbara rekommendationer för att förbättra promptningsupplevelsen. Med en kvalitativ forskningsdesign samlades data in från sex universitetsstudenter genom en workshop och intervjuer.... (More)
Denna kandidatuppsats utforskar hur vardagliga användare, specifikt icke-experter inom högre utbildning, interagerar med AI-baserade chattbottar genom naturliga språkprompter. Tidigare forskning har främst fokuserat på att förbättra modellernas prestanda och förstå promptmekanismer, men har i liten utsträckning uppmärksammat hur användare utan teknisk expertis upplever och formar dessa interaktioner. Studien syftade till att undersöka hur olika sätt att formulera prompter påverkar chattbottens svar och användarnas upplevelse, med målet att generera handlingsbara rekommendationer för att förbättra promptningsupplevelsen. Med en kvalitativ forskningsdesign samlades data in från sex universitetsstudenter genom en workshop och intervjuer. Tematisk analys avslöjade sex centrala fynd som visar att promptformulering inte är en statisk handling utan en dynamisk, iterativ process. Användarna arbetade med trial-and-error, justerade ton och detaljnivå, tilldelade chattbotten roller och använde emotionell eller stilistisk inramning för att styra svaren. Promptning framträdde som en form av aktiv meningsskapande och problemlösning, där användare lärde sig att förfina sina inmatningar för att uppnå bättre resultat. Med stöd i Weicks sensemaking-teori drar studien slutsatsen att användare inte är passiva mottagare utan aktiva medskapare i AI-interaktioner. Uppsatsen ger praktiska rekommendationer såsom att uppmuntra iterativ promptning, meta-promptning, emotionell inramning, målanpassning samt användning av mallar för att stödja ett mer effektivt användande av chattbottar. (Less)
Please use this url to cite or link to this publication:
author
Brown, Kasper ; Elf, Arvid and Henriksson, Edvin LU
supervisor
organization
alternative title
Insights from Everyday User Interactions with AI-based Chatbots
course
SYSK16 20251
year
type
M2 - Bachelor Degree
subject
keywords
Prompt Engineering, AI Chatbots, Human-AI Interaction, Sensemaking, Qualitative Research
language
English
id
9201564
date added to LUP
2025-06-18 08:12:58
date last changed
2025-06-18 08:12:58
@misc{9201564,
  abstract     = {{This bachelor’s thesis explores how everyday users, specifically non-experts in higher education, interact with AI-based chatbots through natural language prompts. While prior research has focused on improving model performance and understanding prompt mechanics, little attention has been given to how users without technical expertise experience and shape these interactions. The study aimed to explore how different ways of phrasing prompts influence chatbot responses and user perceptions, ultimately generating actionable recommendations to enhance the prompting experience. Using a qualitative research design, data were collected from six university students through a workshop and interviews. Thematic analysis revealed six key findings highlighting that prompt formulation is not a static act but a dynamic, iterative process. Users engaged in trial-and-error, adjusted tone and detail, assigned roles to the chatbot, and applied emotional or stylistic framing to guide responses. Prompting emerged as a form of active sensemaking and problem-solving, where users learned to refine inputs to achieve better outcomes. Drawing on Sensemaking theory, the study concludes that users are not passive recipients but active co-creators in AI interactions. The bachelor’s thesis provides practical recommendations such as encouraging iterative prompting, meta-prompting, emotional framing, goal alignment, and the use of templates to support more effective chatbot use.}},
  author       = {{Brown, Kasper and Elf, Arvid and Henriksson, Edvin}},
  language     = {{eng}},
  note         = {{Student Paper}},
  title        = {{The Impact of Prompt Phrasing on AI-based Chatbot Effectiveness and User Experience}},
  year         = {{2025}},
}