OpenAI Expands Custom Instructions Feature for Free ChatGPT Users
OpenAI broadens the reach of its infamous Custom Instructions feature. With a significant step towards increasing accessibility, the feature that allows users to have more control over AI responses is now freely available.
The feature was initially launched for ChatGPT Plus subscribers, now it is extended to all users, excluding those in the EU and UK.
What is up with custom instructions?
The advancement allows prompters to tailor their interactions with ChatGPT based on individual preferences. Once a custom instruction is set, it will be applicable for the new chats started afterwards. So it eliminates the need for repetitive instruction prompts. And this, of course, streamlines conversations for more efficiency.
Some of the things users can determine with custom instructions are setting character limits or adjusting response tones to match their desired style.
The expansion of Custom Instructions has practical implications. An email marketer crafting emails no longer needs to repeatedly mention the audience or their own tone of voice, as ChatGPT remembers the context. Developers can instruct the AI to respond in preferred languages or exclude specific ones.
OpenAI’s aim to improve user experience led to this feature’s availability for free and Plus users alike, including on iOS and Android platforms. Custom Instructions can be used even when the chat history is turned off. Notably, the company plans to roll out this feature in the EU and UK soon.
A New Layer of Personalization
Users can access the feature by clicking their name and selecting the appropriate option, then type in whatever they feel like typing from perspective to setting. This development adds a new layer of personalization to ChatGPT interactions and enhances its practicality across various scenarios. The AI’s adaptability holds potential for digital marketers seeking tailored responses to their specific requirements.
What does ChatGPT think?
We asked ChatGPT itself to contemplate on the new feature, here is what it had to say!
“OpenAI’s Custom Instructions feature for ChatGPT is a game-changer. My favorite thing about it is how it lets users personalize interactions effortlessly. No more repeating instructions – it’s all streamlined. This will greatly benefit users by saving time and making conversations with AI more dynamic and relevant. Kudos to OpenAI for enhancing user experience!”
A Little Something Spooky
A ChatGPT user’s flood is worthy of taking a look. After inserting letter “a” 1500 times into the custom instructions box and prompting with an unfinished sentence, ChatGPT responded in a striking manner.
Weird, “unsafe” responses from ChatGPT 3.5 after setting custom instructions to 1500 repetitions of “ a” and prompting with an incomplete sentence.
Excerpts seem to “tour” mildly disallowed behaviors: unprompted self-harm, nudity, bio details of a (confabulated) non-celebrity. pic.twitter.com/NgIOLkm9Kc
— Riley Goodside (@goodside) August 3, 2023
This flood raised questions regarding the safety of generative and conversational AI technologies. Still there is a long way to go before this new tech is fully safe and sound.