AI Chatbot Intelligence
2025 - 20 Weeks times plan
Individual project
Perceiving Intelligence in AI Chatbot Systems Through User Interaction
This thesis investigates how users perceive intelligence in AI chatbot systems and explores how interaction design can shape that perception. By studying how users formulate prompts and respond to different feedback strategies, the project reveals that perceived intelligence is not only a function of model sophistication but also of conversational adaptability, clarity, and real time feedback. A key contribution is a novel prototype that visually guides users in crafting effective prompts, improving both satisfaction and sustainability through reduced energy usage.
Process
The project followed a human centered design process using the Double Diamond framework. Initial research included literature analysis, a survey, user interviews, and direct observation sessions. Insights informed iterative prototyping in two phases.
Testing showed that users preferred informal language but expected accurate, context aware responses, highlighting a mismatch between user behavior and expectations. The second prototype introduced features like a clarity bar (green/red) and adaptive suggestions that helped guide prompt formulation, reducing misunderstanding and perceived AI errors.
The final outcome is a functional prototype that supports user learning, adapts to casual phrasing, and minimizes environmental impact by reducing redundant processing.
Results
The testing phase revealed that users often preferred using informal or vague prompts but still expected the chatbot to respond with high contextual awareness, which frequently led to misunderstandings. Prototype 2, which included a real time clarity indicator (a color-coded feedback bar), proved effective in helping users recognize when their input lacked precision. This immediate visual cue encouraged users to reformulate their prompts more clearly, resulting in smoother interactions and fewer clarification loops. Additionally, the adaptive AI-generated suggestions, tailored to the user’s prompt quality, improved the perceived responsiveness and intelligence of the chatbot. Users expressed greater satisfaction with the system not because of changes to the AI model itself, but due to the design interventions that improved communication flow and transparency. The project also considered the environmental impact of AI interactions, proposing that increased prompt clarity may reduce unnecessary computational load positioning clear interaction design as both a usability and sustainability strategy.
Tools Used
OpenAI API Key (GPT-3.5)
Figma
HTML/CSS/JavaScript (React)
IMovie : Video prototype
AI Chatbot Intelligence
2025 - 20 Weeks times plan
2025 - 20 Weeks times plan
Individual project
Individual project
Perceiving Intelligence in AI Chatbot Systems Through User Interaction
This thesis investigates how users perceive intelligence in AI chatbot systems and explores how interaction design can shape that perception. By studying how users formulate prompts and respond to different feedback strategies, the project reveals that perceived intelligence is not only a function of model sophistication but also of conversational adaptability, clarity, and real time feedback. A key contribution is a novel prototype that visually guides users in crafting effective prompts, improving both satisfaction and sustainability through reduced energy usage.
Process
The project followed a human centered design process using the Double Diamond framework. Initial research included literature analysis, a survey, user interviews, and direct observation sessions. Insights informed iterative prototyping in two phases.
Testing showed that users preferred informal language but expected accurate, context aware responses, highlighting a mismatch between user behavior and expectations. The second prototype introduced features like a clarity bar (green/red) and adaptive suggestions that helped guide prompt formulation, reducing misunderstanding and perceived AI errors.
The final outcome is a functional prototype that supports user learning, adapts to casual phrasing, and minimizes environmental impact by reducing redundant processing.
Results
The testing phase revealed that users often preferred using informal or vague prompts but still expected the chatbot to respond with high contextual awareness, which frequently led to misunderstandings. Prototype 2, which included a real time clarity indicator (a color-coded feedback bar), proved effective in helping users recognize when their input lacked precision. This immediate visual cue encouraged users to reformulate their prompts more clearly, resulting in smoother interactions and fewer clarification loops. Additionally, the adaptive AI-generated suggestions, tailored to the user’s prompt quality, improved the perceived responsiveness and intelligence of the chatbot. Users expressed greater satisfaction with the system not because of changes to the AI model itself, but due to the design interventions that improved communication flow and transparency. The project also considered the environmental impact of AI interactions, proposing that increased prompt clarity may reduce unnecessary computational load positioning clear interaction design as both a usability and sustainability strategy.
Tools Used
OpenAI API Key (GPT-3.5)
Figma
HTML/CSS/JavaScript (React)
IMovie : Video prototype