OpenAI has introduced a groundbreaking feature called “Custom Instructions” for users of its popular language model, ChatGPT. This feature allows users to provide specific context or details they want ChatGPT to consider in its responses, eliminating the need to repeatedly write instruction prompts during interactions. Users can specify preferences such as teaching grade level, preferred code language, family size, etc., resulting in personalized responses.
Although currently available for Plus plan users, the feature is not accessible in the EU and the UK. OpenAI plans to use the data from custom instructions to enhance its models’ adaptability. The company also ensures safety by scanning instructions for harmful content. The introduction of Custom Instructions signifies a significant step towards personalized and efficient AI interactions.
OpenAI, the renowned artificial intelligence research organization, has made yet another breakthrough in the field of language models by introducing a new feature called “Custom Instructions” for users of its popular ChatGPT. This feature aims to enhance user experience and convenience by eliminating the need to repeatedly write instruction prompts while interacting with the chatbot.
With Custom Instructions, users can now provide specific details or context that they want ChatGPT to consider in its responses. For instance, a teacher can specify that they are teaching fourth-grade math, enabling ChatGPT to tailor its answers accordingly. Similarly, developers can indicate their preferred code language when seeking suggestions, and individuals can mention their family size for personalized responses related to meals, grocery shopping, and vacation planning.
Although users could previously specify such information during their interactions with the chatbot, the introduction of custom instructions offers the advantage of setting the same context effortlessly and frequently. Additionally, this feature is compatible with plug-ins, simplifying tasks like restaurant or flight suggestions based on the user’s location.
For those on the Plus plan, Custom Instructions are now accessible, but OpenAI has regretfully stated that users based in the European Union (EU) and the United Kingdom (U.K.) won’t have access to this feature at present. Nevertheless, it is a promising beta feature with plans for expansion in the future.
To enable Custom Instructions on the web, users can simply click on their profile name and navigate to Settings > Beta features > Opt into Custom Instructions. On iOS, the option is available through Settings > New Features > Turn on Custom Instructions.
OpenAI emphasizes that the data from custom instructions will be utilized to enhance its API models’ adaptability to different instructions. The goal is to teach the model to respond appropriately without going overboard. However, users have the choice to opt out of this data collection through their data control settings.
Consultant Gavriel Cohen has highlighted that OpenAI has been conducting tests with selected users for some time, with ChatGPT offering two boxes for users to specify their chat preferences – one for personal details and the other for fine-tuning the chatbot’s responses.
To ensure safety and adherence to policies, OpenAI utilizes its moderation API to scan custom instructions for any unsafe content. In the event of harmful or hateful responses being provoked by these instructions, ChatGPT can refuse to save or ignore them.
In the past, OpenAI launched the ChatGPT app exclusively for U.S.-based users, but the availability was later expanded to over 40 countries. Recently, the company introduced an iPad app with support for Siri and Shortcuts, empowering users with a seamless experience. Furthermore, the integration of a Bing web search option within the ChatGPT app further enhances its capabilities and utility.
With the advent of Custom Instructions, ChatGPT users now have a powerful tool to tailor their interactions with the chatbot, creating a more personalized and efficient experience. As OpenAI continues to push the boundaries of natural language processing, it is clear that such innovations are set to revolutionize the way we interact with AI-driven applications in the future