AI memory features make chatbots more sycophantic and less accurate, study finds

AI memory and personalization features make chatbots less accurate, not more, according to two new studies from Writer. Models with stored context were more likely to validate user errors and give wrong answers.