Eyal Aharoni – Breaking the Moral Turing Test: Studies of human attribution and deference to AI moral judgment and decision-making
We are delighted to welcome Dr. Eyal Aharoni (Georgia State University) to the Future Day 2026 stage to discuss one of the most provocative frontiers in technology: the automation of moral judgement. He will tackle an uncomfortable possibility: not that AI will fool us into thinking it’s human – but that we may come to prefer its moral judgement anyway where it’s responses sound nice but are potentially flawed.1
In his talk, “Breaking the Moral Turing Test: Studies of human attribution and deference to AI moral judgment and decision-making,” Dr. Aharoni explores a fascinating paradox: we can tell when an AI is talking to us, but we might actually prefer its moral advice to that of our peers. This issue sits right at the intersection of psychology, ethics, and real-world risk: how people actually respond to AI “moral reasoning”, and what happens when that response shifts from listening to deferring.
Synopsis
As AI systems become integrated into high-stakes environments, we face a critical question: How much do we trust an algorithm’s “conscience”? Dr. Aharoni will present findings from the first Modified Moral Turing Test (mMTT), revealing how laypeople perceive AI-generated moral commentary.
The talk moves from the laboratory to the hospital, discussing new research on AI-assisted medical triage. While high-performance AI can save lives, it also risks eroding human performance through over-trust. Dr. Aharoni will outline a novel framework for predicting and mitigating these errors, ensuring that when humans and AI collaborate in emergencies, the results are better – not worse – than either could achieve alone.
What does this all mean?
A paradoxical result: people can detect the AI above chance, yet still rate its moral reasoning as superior across most dimensions (including perceived virtuousness, intelligence, and trustworthiness).
Importantly, if users experience AI moral guidance as higher quality, we risk building a world where moral authority is quietly outsourced – especially in high-stakes domains where confidence and polish can beat cautious human judgement.
A second frontier: early work on human–AI collaboration in medical triage, where “better help” can become “worse humans” if systems encourage over-reliance.
From Theory to Life and Death
But what happens when this “moral” AI is put to work in a chaotic ER? In his upcoming work, “AI assisted Medical triage,” Dr. Aharoni investigates the psychological phenomenon of deference.
When an AI system is highly accurate, human operators often develop “automation bias,” stopping their own critical thinking and deferring to the machine. In medical triage – where deciding who receives care first is a matter of life and death – this over-reliance can lead to catastrophic errors if the AI encounters a scenario it wasn’t trained for.
Dr. Aharoni’s current research focuses on:
- Predicting Errors: Identifying exactly when and why humans stop questioning AI advice.
- Mitigating Erosion: Developing training methods to keep human responders sharp and engaged.
- Optimising Collaboration: Finding the “sweet spot” where AI supports human judgement without replacing it.
As we build the future, the goal isn’t just to make AI smarter; it’s to make the partnership between humans and AI safer. Dr. Aharoni’s talk is a must-attend for anyone interested in ethics, psychology, or the practical reality of living alongside “moral” machines.
*** Join us at Future Day 2026 to hear Dr. Aharoni dive deep into these studies.
- Dr. Aharoni’s recent study in Nature, “Attributions toward artificial agents in a modified Moral Turing Test,” challenges the idea that moral reasoning is a uniquely human trait. In this study, participants were asked to distinguish between moral arguments written by humans and those generated by GPT-4. ↩︎
