GPT-4 and GPT-3.5 Have Problems With Hallucinations

The OpenAI team had both GPT-4 and. GPT-3.5 take a bunch of exams, including the SATs,. The tests were designed to test their ability on real world problems such as reading comprehension or problem solving. But they also found issues with biased training data that can lead them to make false claims about what is going on around them. They say we need to be more transparent about how hallucinations affect the trustworthiness of models like G PT-4.