Researchers Warn Human-Like AI may Shift Blame from Real Decision-Makers
When AIs appear more human-like, people assign them greater blame for moral harm, potentially reducing accountability for human stakeholders.
A new study published in PLOS ONE suggests that perceptions of artificial intelligence (AI) as having human-like minds lead people to assign the technology greater blame for moral transgressions. Conducted by Minjoo Joo of Sookmyung Women’s University, Seoul, the research examined how participants judged scenarios where AI systems were involved in real-world harm, such as discriminatory image tagging.
The experiments presented participants with transgressions by AIs and then asked them to evaluate blame distribution among different agents, including the AI’s programmers and associated companies. By manipulating AI characteristics—giving them human-like names, ages, heights and hobbies—researchers were able to measure how changes in perceived AI “mind” attributes affected blame judgements.
Reduced Corporate Blame With More Human-Like AI
The findings indicate that participants assigned significantly more blame to AIs when these systems were perceived as having human-like minds. In scenarios where the AI seemed more “aware” or “capable of feeling”, people were less inclined to hold the company responsible. As Joo notes, a key effect of viewing AIs as human-like was “reducing blame on human stakeholders”. However, when asked to judge blame independently for each agent rather than comparatively, the level of blame against the company did not fall.
This subtle shift in blame attribution could have considerable implications. Organisations behind AI technology might benefit from perceptions of AI sentience, intentionality and independent decision-making—allowing them to sidestep scrutiny. Over time, if this tendency becomes widespread, accountability for AI-driven harm could slip through the cracks.
Caution Against Moral Scapegoating
While these findings may prompt developers and institutions to emphasise their AI’s supposed humanity—touting advanced cognition or empathy—the study’s author urges caution. Falsely attributing blame to AI risks encouraging its misuse as a scapegoat, potentially undermining accountability in critical fields such as healthcare, military operations and financial decision-making.
“Can AIs be held accountable for moral transgressions? This research shows that perceiving AI as human-like increases blame toward AI while reducing blame on human stakeholders, raising concerns about using AI as a moral scapegoat,” says Joo.
Ultimately, while AIs are becoming more sophisticated, they remain tools—no matter how human they might seem. This study highlights the importance of recognising that responsibility still lies with the humans who design, deploy and control AI systems. By maintaining a clear-eyed view of AI as a clever instrument rather than an autonomous moral agent, we ensure accountability stays where it belongs.
