Large language models (LLMs), the artificial intelligence (AI) systems underpinning ChatGPT and similar conversational platforms, are now used by many people worldwide to find and summarize information and generate different types of text. Despite their widespread use, these models still have notable limitations.

When generating text or answers to user queries, current LLMs do not rely only on patterns observed and data analyzed during training. They can also retrieve information from external sources, such as websites, databases and search engines.

Giving LLMs access to external resources allows them to produce responses that are up to date and more comprehensive. If a model cannot reliably judge the reliability of external sources of information, however, it may generate text that is untrustworthy or inaccurate.