Three Years On, How Accurate Are AI Chatbots?

ChatGPT's 2022 launch significantly popularized AI chatbots, also known as large language models (LLMs). LLMs have advanced in processing complex information and aiding tasks like coding. However, they still face challenges. Accuracy remains a problem, with hallucinations and misinterpretations persisting in these models. A study from the European Broadcasting Union and the BBC found many responses contained inaccuracies. Almost half of responses from popular chatbots showed accuracy problems in a study from May-June 2025. Significant errors, like sourcing issues, were also observed. A previous study in December 2024 showed an even higher rate of inaccurate responses. These inaccuracies raise concerns about reliability. These issues are particularly concerning in applications like healthcare and education. Users must understand these limitations as AI development continues at a rapid pace.