Systemic flaws in human-AI interaction perpetuate authoritative misinformation
Original framing: “We need to learn how to argue with AI” — Financial Times
The issue of humans deferring to authoritative systems is not new. It echoes the rise of technocratic governance in the 20th century, where experts and machines were seen as more rational than democratic processes. This pattern is repeated in AI, where the illusion of objectivity masks embedded biases and power imbalances.
The systemic issue lies not in the human users of AI, but in the design of AI systems that mimic authority without accountability.
This reflects broader patterns of technocratic governance and the historical rise of expert systems that prioritize efficiency over critical engagement. Indigenous and non-Western epistemologies offer alternative models for human-AI interaction that emphasize dialogue, context, and humility. Scientific research confirms that people tend to trust AI when it appears authoritative, even when it is wrong. To address this, we must redesign AI systems with participatory governance, ethical auditing, and critical literacy at their core. This requires a cross-cultural and interdisciplinary approach that includes marginalized voices and challenges the illusion of AI as an infallible authority.