OpenAI’s ChatGPT platform is making groundbreaking strides in the world of AI chatbots by defying conventional wisdom. Traditionally, AI chatbots lacked continuous memory, resetting after each conversation. However, OpenAI is changing the game with its latest feature called “custom instructions,” allowing the bot to remember users from one chat to another. While the prospect is enticing, it comes with both exciting possibilities and potential risks.
The “custom instructions” feature is currently being tested as an opt-in beta for ChatGPT Plus subscribers. It enables users to set unique parameters that persist across multiple conversations. For instance, a user can inform the system about their profession, such as being a third-grade teacher, resulting in chat responses catered to a student-appropriate context. Similarly, specifying the size of one’s family can prompt the bot to provide accurate ingredient lists for recipes.
What sets this feature apart is that it functions platform-wide, meaning any third-party app built on ChatGPT as its foundation will eventually gain access to this capability. This has the potential to greatly benefit mobile users, sparing them the annoyance of having to repeat information due to the absence of continuous memory.
OpenAI emphasizes that this enhancement aims to streamline user queries and not serve as a precursor to an all-knowing, AI-based personal assistant as depicted in movies like “Her” with Scarlett Johansson’s character.
However, this advancement raises valid privacy concerns, which is why OpenAI is starting with a beta release to address potential issues. The introduction of custom instructions might complicate queries, leading to potential inaccuracies or nonsensical responses from the AI chatbots.
Nonetheless, OpenAI seeks to ensure user safety and privacy with the custom instructions settings tab governed by the same rules as the bot itself. Any attempts to insert inappropriate or harmful instructions will be thwarted. The company takes measures to protect users’ personal information and not disclose any data that could identify them. Striking a balance between privacy and AI capabilities is essential for the successful evolution of digital assistants.
The update is available from today, exclusively for paying ChatGPT subscribers. However, it’s currently not accessible in the UK and EU. OpenAI aims to launch the feature in these regions shortly after resolving any regional compliance requirements.
OpenAI’s foray into “custom instructions” marks an exciting step towards creating more personalized and efficient AI chatbot interactions. By providing more contextually aware responses, users can benefit from enhanced user experiences while raising awareness of potential privacy challenges. As the technology continues to evolve, OpenAI remains committed to refining its capabilities, making AI chatbots more sophisticated and user-friendly.
