AI-modellen als vervanger voor enquêtedeelnemers: wat onderzoek toont

Researchers are exploring whether large language models can substitute for human survey participants by addressing a key technical challenge: mode collapse, where AI systems generate repetitive responses. Unlearning techniques show promise in improving the diversity and authenticity of synthetic survey data, potentially offering a cost-effective alternative to traditional respondent recruitment.