Ensembling vs. Delegating: Different Types of AI-Involved Decision-Making and Their Effects on Procedural Fairness Perceptions
Christopher Diebel, Akylzhan Kassymova, Mari-Klara Stein, Martin Adam, and Alexander Benlian
This study investigates how employees perceive the fairness of decisions that involve artificial intelligence (AI). Using an online experiment with 79 participants, researchers compared scenarios where a performance evaluation was conducted by a manager alone, fully delegated to an AI, or made by a manager and an AI working together as an 'ensemble'.
Problem
As companies increasingly use AI for important workplace decisions like hiring and performance reviews, it's crucial to understand how employees react. Prior research suggests that AI-driven decisions can be perceived as unfair, but it was unclear how different methods of AI integration—specifically, fully handing over a decision to AI versus a collaborative human-AI approach—affect employee perceptions of fairness and their trust in management.
Outcome
- Decisions fully delegated to an AI are perceived as significantly less fair than decisions made solely by a human manager. - This perceived unfairness in AI-delegated decisions leads to a lower level of trust in the manager who made the delegation. - Importantly, these negative effects on fairness and trust do not occur when a human-AI 'ensemble' method is used, where both the manager and the AI are equally involved in the decision-making process.
Host: Welcome to A.I.S. Insights, the podcast where we turn complex research into actionable business intelligence. I’m your host, Anna Ivy Summers. Host: Today, we're diving into a fascinating study titled "Ensembling vs. Delegating: Different Types of AI-Involved Decision-Making and Their Effects on Procedural Fairness Perceptions". Host: It’s all about how your employees really feel when AI is involved in crucial decisions, like their performance reviews. And to help us unpack this, we have our lead analyst, Alex Ian Sutherland. Welcome, Alex. Expert: Thanks for having me, Anna. It’s a critical topic. Host: Absolutely. So, let's start with the big picture. What's the core problem this study is trying to solve for businesses? Expert: The problem is that as companies rush to adopt AI for HR tasks like hiring or evaluations, they often overlook the human element. We know from prior research that decisions made by AI can be perceived by employees as unfair. Host: And that feeling of unfairness has real consequences, right? Expert: Exactly. It can lead to a drop in trust, not just in the technology, but in the manager who chose to use it. The study points out that when employees distrust their manager, their performance can suffer, and they're more likely to leave the organization. The question was, does *how* you use the AI make a difference? Host: So how did the researchers figure that out? What was their approach? Expert: They ran an online experiment using realistic workplace scenarios. Participants were asked to imagine they were an employee receiving a performance evaluation and their annual bonus. Expert: Then, they were presented with three different ways that decision was made. First, by a human manager alone. Second, the decision was fully delegated by the manager to an AI system. And third, what they call an 'ensemble' approach. Host: An 'ensemble'? What does that look like in practice? Expert: It’s a collaborative method. In the scenario, both the human manager and the AI system conducted the performance evaluation independently. Their two scores were then averaged to produce the final result. So it’s a partnership, not a hand-off. Host: A partnership. I like that. So after running these scenarios, what did they find? What was the big takeaway? Expert: The results were incredibly clear. When the decision was fully delegated to the AI, participants perceived the process as significantly less fair than when the manager made the decision alone. Host: And I imagine that had a knock-on effect on trust? Expert: A big one. That perception of unfairness directly led to a lower level of trust in the manager who delegated the task. It seems employees see it as the manager shirking their responsibility. Host: But what about that third option, the 'ensemble' or partnership approach? Expert: That’s the most important finding. When the human-AI ensemble was used, those negative effects on fairness and trust completely disappeared. People felt the process was just as fair as a decision made by a human alone. Host: So, Alex, this is the key question for our listeners. What does this mean for business leaders? What's the actionable insight here? Expert: The main takeaway is this: don't just delegate, collaborate. If you’re integrating AI into decision-making processes that affect your people, the 'ensemble' model is the way to go. Involving a human in the final judgment maintains a sense of procedural fairness that is crucial for employee trust. Host: So it's about keeping the human in the loop. Expert: Precisely. The study suggests that even if you have to use a more delegated AI model for efficiency, transparency is paramount. You need to explain how the AI works, provide clear channels for feedback, and position the AI as a support tool, not a replacement for human judgment. Host: Is there anything else that surprised you? Expert: Yes. The outcome of the decision—whether the employee got a high bonus or a low one—didn't change how they felt about the process. Even when the AI-delegated decision resulted in a good outcome, people still saw the process as unfair. It proves that for your employees, *how* a decision is made can be just as important as the decision itself. Host: That is a powerful insight. So, let’s summarize for everyone listening. Host: First, fully handing off important HR decisions to an AI can seriously damage employee trust and their perception of fairness. Host: Second, a collaborative, or 'ensemble,' approach, where a manager and an AI work together, is received much more positively and avoids those negative impacts. Host: And finally, a good outcome doesn't fix a bad process. Getting the process right is essential. Host: Alex, thank you so much for breaking that down for us. Incredibly valuable insights. Expert: My pleasure, Anna. Host: And thank you to our audience for tuning in to A.I.S. Insights, powered by Living Knowledge. We’ll see you next time.
Decision-Making, Al Systems, Procedural Fairness, Ensemble, Delegation
