OpenAI has updated its ChatGPT feature to include a new "reference chat history" option, which allows the AI to use all prior conversations as context and adapt future responses accordingly. This is in addition to the existing "memory" feature, which allows ChatGPT to retain a limited number of pieces of information for future responses. The new feature enables ChatGPT to provide more personalized and accurate responses based on a user's conversation history. In a test, the author asked ChatGPT to describe them based on all their chats, and the AI provided a detailed and accurate response, highlighting the user's analytical nature, technical literacy, and interests. In contrast, when the author asked Siri to do the same, it responded with an error message. The update is a significant improvement over the previous "memory" feature, which required users to specifically ask ChatGPT to remember certain information. The new feature is automatically enabled, unless users choose the "Temporary Chat" option, which is similar to incognito mode. The update demonstrates the potential of machine learning to lead to incremental improvements in products, eventually adding up to major gains. However, not everyone is convinced of the value of chatbots, including John Giannandrea, who has expressed a dim view of their usefulness in the past. Despite this, the new feature is a significant step forward for ChatGPT and has the potential to revolutionize the way we interact with AI.
arstechnica.com
arstechnica.com
