Artificial intelligence cannot replace people in opinion polls yet, as AI-generated synthetic respondents give answers that differ significantly from real human responses.

Analysis

People might believe AI can easily simulate public opinion because large language models are skilled at mimicking human language and processing vast amounts of demographic data quickly. However, this perception misses the nuanced, dynamic nature of human beliefs, which are constantly shaped by current events, personal experiences, and emotional shifts that static training data cannot fully capture. Different AI models can also skew data in opposite directions—such as pushing answers to extremes or softening them toward the middle—making them unreliable for capturing true public sentiment. This matters greatly to readers because relying on synthetic polling could lead to deeply flawed political strategies, misinformed news reporting, and a fundamental loss of public trust in democratic data collection.