Both papers found that chatbots influence voter opinions not by using emotional appeals or storytelling, but by flooding the user with information. The more information the chatbots provided, the more persuasive they were — but they were also more likely to produce false statements, the authors found.