OpenAI’s ChatGPT app has received an exciting new update that introduces a basic memory feature called custom instructions. This feature allows users to set their preferences and influence how ChatGPT responds in conversations, providing a semblance of memory to the AI system.
OpenAI took user feedback into account, as many expressed dissatisfaction with starting each chat from scratch. With this new update, users can now set specific instructions that ChatGPT will remember, eliminating the need for repetitive instructions.
For instance, a data scientist can instruct ChatGPT to remember that they are working with a particular dataset and need assistance with data cleaning and outcome prediction. An author writing a science fiction novel can specify details about the genre, setting, characters, and logline, allowing ChatGPT’s suggestions to be tailored accordingly.
Moreover, this update is valuable for individuals like crypto traders, who may want to discuss the latest trends in Bitcoin’s price. By setting the AI to remember their interest in Bitcoin, any future market analysis generated by ChatGPT will revolve around Bitcoin, eliminating the need to specify their interest repeatedly.
While the custom instructions feature is currently in beta, it already shows promise in enhancing the use of plugins by offering contextual information that can be applied across different plugins. This means users interested in analyzing data, such as charts and price movements, can have plugins fetch the relevant information without having to request it every time.
OpenAI has also addressed privacy concerns associated with this update. The company ensures that any instructions violating their Usage Policies are not stored. Additionally, steps have been taken to remove personal identifiers found in custom instructions, improving model performance.
However, despite these positive advancements, concerns have been raised about the performance of GPT-4, the underlying technology of ChatGPT Plus. A recent research paper from UC Berkeley and Stanford University compared GPT-4 to its predecessor, GPT-3.5, and revealed a decline in performance over time.
While the integration of a basic memory feature is a significant step forward, the true potential of this update will only be revealed as users continue to explore and interact with this enhanced tool.
It’s worth noting that this article adheres to guidelines in providing an SEO-friendly and conversational tone, refraining from promotional content. The journalistic integrity of this piece is maintained by presenting different perspectives and opinions where relevant.
