OpenAI, the most popular leader in artificial intelligence, has introduced custom instructions on ChatGPT. This feature is currently in a beta program so only select users can use it.
With this, users will no longer be required to input the same prompts all over again every time they access the AI chatbot.
As The Verge reports, custom instructions are here to store the usual preferences of the users.
ChatGPT Custom Instructions: What Does it Do?
As OpenAI writes, the new ChatGPT feature focuses on remembering the user's life story and other information that the bot wants to know from you.
In this regard, those people who frequently forget the prompts they use will now have no problem remembering them the next they open the tool.
For instance, if you enter a prompt about programming, ChatGPT can specify your instructions for your next request. It's simply good since your instructions can be stored.
At the same time, if you always write articles and use the same prompts daily, this feature comes to play in an instant. Since it already knows a similar context, it's easier to inform you when it comes to suggestions.
For those who are curious if it works with plugins, the answer is yes. Just in case you ask ChatGPT regarding a decent restaurant nearby your area, it will right away hand you the prompt it last recalled.
Related Article : Can Apple GPT Compete with ChatGPT? Gurman Says Generative AI Features Coming in 2024
Are Custom Instructions Available to Everyone?
TechCrunch reports that since it's a beta feature at the time, only Plus subscribers in the UK and EU are eligible to access the feature. Some other time, we might see OpenAI rolling it out in other regions.
"Information from your use of custom instructions will also be used to improve model performance - like teaching the model how to adapt its responses to your instructions without overdoing it," the company said.
If you wish to stop using the feature, you can go to the data control settings and opt it out.
OpenAI says that there's a limitation to the responses on the platform. The answers won't exceed 1,500 characters, according to the AI bot maker.
If the customized instructions are deemed unsafe, OpenAI can use its moderation API. Of course, it has the power to ignore or enable the instructions depending if it will violate the existing policy of the company.
This way, users can't abuse the chatbot and they can't spam hateful responses on the platform.
As per Joanne Jang, who is working on OpenAI's model behaviors, the chatbot does not know the user who uses it. Once a new thread opens, ChatGPT will automatically forget everything about you, but it can recall other things that might have been used in other conversations.
Meanwhile, Tech Times reports that ChatGPT and other large language models can flag fake news. However, they are still not as reliable as human fact-checkers.
