Overview

The thesis will investigate how AI-based assistants can support idea evaluation, concept refinement, and selection during the early stages of innovation and entrepreneurship. While generative AI tools such as ChatGPT are increasingly used to generate large numbers of ideas, much less is known about how these systems can help innovators converge from many raw ideas toward a smaller set of well-developed, viable concepts. This project aims to examine how AI-based assistants can act as evaluators, critics, and refinement partners, and how different forms of AI-supported evaluation influence concept quality, decision confidence, and user experience in startup contexts.

Background

Recent advances in generative AI have enabled widespread access to powerful language models such as ChatGPT, Claude, or Gemini. These systems are increasingly embedded in entrepreneurial and innovation-related work, particularly in ideation and early concept development. Prior research in information systems and entrepreneurship suggests that human–AI teams can outperform humans alone in creative tasks such as idea generation. As a result, early-stage innovators can now generate a large number of ideas quickly and at low cost.

However, innovation processes are not only characterized by divergence but also by convergence. After ideation, entrepreneurs must evaluate ideas, select promising concepts, and refine them into more detailed and coherent opportunity descriptions. These tasks involve judgment, comparison, and decision-making under uncertainty and are often more challenging than idea generation itself. Generative AI introduces new opportunities in this phase, for example by providing structured feedback, highlighting strengths and weaknesses, suggesting improvements, or comparing ideas against evaluation criteria. At the same time, AI-generated feedback may influence human judgment in subtle ways, raising questions about trust, over-reliance, and responsibility.

Despite growing interest in AI-supported creativity, existing research has largely focused on idea generation. We still lack systematic empirical evidence on how AI-based assistants can support idea evaluation, concept refinement, and selection, and how such support affects decision quality, perceived confidence, and ownership in early-stage innovation. This represents a critical gap in information systems research on AI-based decision support and knowledge augmentation in entrepreneurial contexts.

Research Objectives

The overarching goal of this master thesis is to investigate how AI-based assistants can be designed to support concept refinement and evaluation in early-stage innovation. Rather than focusing on idea generation, the thesis explicitly addresses the convergent phase of the innovation process, in which rough ideas are elaborated, critically assessed, and transformed into more mature and communicable innovation concepts.

The thesis will design and empirically evaluate an AI-assisted concept refinement and evaluation system that supports users in systematically improving early-stage ideas. The AI-based assistant may provide structured feedback (e.g., strengths and weaknesses), suggest concrete refinements, highlight missing elements, or guide users through predefined evaluation criteria relevant to startup contexts (e.g., problem clarity, value proposition, feasibility, or differentiation).

Your Profile

• You are interested in the emerging field of generative AI

• You are interested in innovation, entrepreneurship, and decision-making

• You enjoy empirical research (experiments, surveys, data analysis)

• You have experience with Python or are willing to learn

• You are highly motivated to work in a self-organized and goal-oriented manner and bring in your own ideas

• Very good English skills, as the thesis will be written in English

We offer an exciting research topic with strong relevance to both academia and practice, close supervision, and the opportunity to develop theoretical, methodological, and practical skills. If you are interested, please send a current transcript of records, a short CV, and a brief motivation (2–3 sentences) to Jonas Liebschner (jonas.liebschner∂kit.edu).