The role of transparency in AI systems on human idea evaluation processes

• Main Author: Hessing, Nicolai
• Publication Date: 2023
• Format: Master thesis
• Language: eng
• Source: Repositórios Científicos de Acesso Aberto de Portugal (RCAAP)
• Download full: http://hdl.handle.net/10400.14/43427
• topic: Artificial Intelligence; Transparency; Idea evaluation; Judge-advisor system; Inteligência Artificial; Transparência; Avaliação de ideias; Paradigma JAS

he evaluation of innovative ideas is a key process for successful new product development (NPD), which traditionally relies on human input. The rise of big data has introduced artificial intelligence (AI) as a new promising source of idea evaluation. Characterized by enhanced analytical power, heightened predictive accuracy, and advanced learning capabilities, AI emerges as an efficient alternative to human intelligence and has the potential to revolutionize the entire innovation process. Nonetheless, the intricate nature of AI systems also raises concerns about their trustworthiness and accountability, thereby underscoring the need for enhanced transparency. However, there is ambiguity concerning the potential implications of enhanced transparency in AI systems for human decision-making processes (e.g., idea evaluations). This thesis investigates the impact of enhanced transparency in AI systems on entrepreneurs' inclination to utilize AI advice within the context of idea evaluation processes. Furthermore, potential moderating effects of confidence and familiarity with AI are examined. Data derived from an experimental survey encompassing 366 entrepreneurs is examined employing the Judge-Advisor System (JAS). The findings reveal a significant positive relationship between the degree of transparency in AI systems and AI advice utilization. The relationship is moderated moderately by the level of familiarity with AI and surprisingly not moderated by the level of confidence in the entrepreneurs' own evaluation.