Samenvatting

The rapid advancement of Generative AI (GenAI), particularly large language models (LLMs), has significantly reshaped how individuals and organizations engage in decision-making. From customer service and education to strategic planning and design, these technologies are no longer limited to providing passive recommendations. Instead, they are increasingly being integrated into workflows to co-construct solutions with users. This evolution raises important questions about how different AI modalities influence the quality of decision outcomes and user experience across various phases of decision-making.

The article highlights that delegative AI represents automation—fast, efficient, and low-effort—where users offload cognitive work.
Conversational AI, by contrast, represents collaboration—a back-and-forth process that stimulates reflection and refinement.
Unlike many studies focusing on AI’s accuracy or fairness, this work looks at how people experience working with AI differently depending on whether the system acts as a delegate or as a conversation partner.

The key innovation is applying psychological and cognitive theories to explain these effects. We combined Cognitive Fit Theory and Dual-Process Theory to show that conversational AI better aligns with how humans think and reason during decision-making.