Artificial intelligence (AI) is playing an increasingly significant role in decision-making processes across various domains, from financial services to healthcare and resource allocation. While AI-based systems promise benefits such as improved efficiency, consistency, and scalability, their integration into decision-making processes is dependent on their perceived fairness and users’ trust. Here, we use a vignette-based experimental approach comparing several decision scenarios to assess how differences in task type (human vs. mechanical) impact on how trustworthy and fair participants’ perceive corresponding AI-based decision-support systems to be. Human-centric tasks involve social or emotional judgment (e.g., evaluating a candidate's team fit), while mechanical tasks are rule-based and data-driven (e.g., optimizing warehouse task allocation). Importantly, we further contrast AI-based decisions with decisions presumably made by a human based on the same information. Understanding how the perception and evaluation of AI-based versus human decisions changes based on task type is essential to derive corresponding suggestions on how to increase the acceptance and adoption of AI-based decision support.