HUMAINE, a large-scale UK-US study, found that health and wellbeing, not politics, dominate how people interact with AI. Across 40,000 conversations, users frequently discussed fitness, mental health, and medical topics, often viewing AI as a trusted confidant or health advisor. However, current evaluation methods for artificial intelligence focus too heavily on technical metrics, overlooking the human impact.
Researchers warn that this creates dangerous blind spots, especially in sensitive areas such as mental health. They call for practical, scenario-based, and human-centered testing to ensure AI supports mindful collaboration rather than mindless automation. True progress, the study concludes, lies in balanced, reliable, and ethically aware systems.