Work

Vaatimustenmäärittelykäytännöt tekoälysysteemeille – Tapaustutkimus keskikokoisessa tekoälyn konsultointiyrityksessä

Year: 2026
Type: dissertation
Abstract: AI has rapidly become a priority investment across different industries due to its promises to improve efficiency and reduce costs. Requirements engineering (RE) is an important step of software development and has been studied for several decades, but its application in AI projects is still a less mature research direction. Establishing sound RE practices for AI development is critical to ensure high-quality, reliable AI systems. The goal of this thesis was to investigate what kind of a requirements engineering process supports the development of AI systems. The key challenges of RE in AI development were also identified. The empirical part of the study was conducted as a case study in a mid-sized AI consulting company. The case study consisted of 7 semi-structured interviews, and the interviewees ranged from less technical roles to highly technical roles. This ensured diverse viewpoints, complementing each other. The research shows that AI development often suffers from insufficient data requirements and large knowledge gaps between stakeholders. The case company also had challenges regarding requirements documentation. Both the literature review and the empirical study supported the findings that multidisciplinary collaboration, comprehensive data requirements and trade-off analysis for AI-specific requirements are essential in requirements engineering for AI systems. This study concludes that neglecting data-related requirements can result in poor quality AI systems and even project failure. Additionally, insufficient quality requirements can also create challenges in the development process and cause excessive iterations in the project.
Institution: Aalto University
Language: English
Open Access status: green
OpenAlex ID: W7204015994
Cites: 57
Cited by:
Related to: 10