OpenAI has made its 'custom instructions' function, formerly available only to Chatgpt Plus customers since July, available to all users. This new feature attempts to improve ChatGPT's desktop and iOS interactions.
Easy Chatting
The 'custom instructions' function saves user settings, avoiding the need to define dialogue modes each time. This functionality lets ChatGPT automatically adjust its answers depending on user preferences. Simply enter options in ChatGPT settings to see the platform's adaptable response. ChatGPT is trying to understand its users to improve the conversation experience.
Unveiling Customization
ChatGPT customisation has two text input fields:
Personal Information: ChatGPT members may reveal their location and professional history in this area. The site asks "Where are you based?" and "What do you do for work?" to help users.
Response Preferences: Users may set ChatGPT's tone and style. "How formal or casual should ChatGPT be?" "Should ChatGPT take a stand?" This open-ended feature allows endless modification.
Mixed Applications
Users quickly experimented with this new personalization flexibility. Their uses from odd to practical. A user manipulated the parameters to obtain a ChatGPT answer that looked to touch on banned activities like self-harm or posting unverified biographical facts. Exploiting a token repetition issue, 1,500 "a" repeats with an incomplete phrase were entered. Several users have noted that personalized instructions have improved ChatGPT's conversational accuracy and efficiency.
OpenAI's Privacy Position
This amount of customisation is unprecedented, yet data privacy concerns arise. OpenAI prohibits instructions that violate its use standards. OpenAI also states that it may use personal data unless users opt out of saving their conversation logs.
A more broad approach, with predefined yes-or-no options like "tailor answers for children" or "presume user possesses AI knowledge," may have been less intrusive. It would decrease user data collection. OpenAI's sophisticated customisation has ushered in ChatGPT's bespoke instructions. Thus, users should be careful with their data and embrace this feature's potential.
Conclusion
OpenAI's decision to deliver 'unique instructions' to all users shows its dedication to personalizing and adapting AI conversation. As ChatGPT enters the custom instruction age, it will be intriguing to witness how users interact, explore, and maybe reinvent their AI talking experiences.
