Experts raise concerns over proposals to use AI in human research
• Recent proposals suggest using AI to generate research data instead of human participants, citing benefits like reduced costs and risks.
• But experts worry this could produce poor quality, biased results that fail to capture human complexity.
• AI models lack lived experiences and emulate humans by aggregating diverse data, potentially perpetuating stereotypes.
• Using AI to study vulnerable groups is very risky and could erroneously inform future policies.
• International guidelines are needed on acceptable use of AI in research to prevent undermining quality.