Friday, 24 March, 2023

OpenAI Confirms Source of ChatGPT's History Mix-Up

<p>OpenAI has confirmed that the source of the historical inaccuracies found in its language model, ChatGPT, was due to an error in the training data used by the model. The error occurred when a portion of the data used to train ChatGPT was inadvertently mixed with historical fiction, leading to the model generating inaccurate historical facts and events. OpenAI has since removed the affected data and retrained the model to ensure the accuracy&nbsp;of its responses.</p>