OpenAI Adds Much Needed Custom Instructions Memory Function to Save You Tons of Time

OpenAI has enhanced its chatbot, ChatGPT, to be more time-saving by adding a feature that makes it retain user preferences. The new feature allows users to give ChatGPT personalized instructions on how its responses should be tailored to fit the user’s precise needs.

OpenAI Makes ChatGPT a Better Assistant

Since its launch, ChatGPT has been a work in progress with new features being added to make it better and also retain its leading position amidst competing chatbots. The company behind the chatbot has continued to train its models, provide a more powerful premium version, and even offered a mobile app.

Until now, users have had to provide very long prompts in which they include specifications and context every single time to get responses that are more tailormade to their situation. However, in its latest update, the chatbot is being made a better assistant.

ChatGPT just dropped a game-changing new feature:

Custom Instructions.

Here’s why this new update will make ChatGPT an even better personal assistant: pic.twitter.com/A9bNvZ31AO

— Alex Brogan (@_alexbrogan) July 21, 2023

Responding to user feedback, the Sam Altman-led company said:

“We’ve heard your feedback about the friction of starting each ChatGPT conversation afresh. Through our conversations with users across 22 countries, we’ve deepened our understanding of the essential role steerability plays in enabling our models to effectively reflect the diverse contexts and unique needs of each person.”

OpenAI has created a feature named custom instructions which is meant to improve the quality of responses ChatGPT provides based on the user’s preferences. Through this feature, users provide ChatGPT with a list of things they would like for the chatbot to consider as it searches for and generates its answer.

For instance, the company says that “a teacher crafting a lesson plan no longer has to repeat that they’re teaching 3rd-grade science. A developer preferring efficient code in a language that’s not Python – they can say it once, and it’s understood.”

A user could also tell the chatbot the size of their family so that whenever they ask questions concerning meal plans, grocery lists, or even vacation planning, ChatGPT will provide answers that suit the family size.

The new feature is designed to save users from repeating certain information over and over again when giving ChatGPT prompts. If a user is on vacation and they state their location as Paris, for instance, every time they ask for recommendations of activities, hotels, or flights, the user will receive Paris-specific responses without repeating their location each time.

The feature will likely prove to be incredibly useful for professionals using ChatGPT because they constantly have to repeat the same basic context and information in every prompt. Setting customer instructions could save a tremendous amount of time over many uses.

Personalize Your OpenAI ChatGPT Experience

Custom instructions have been made optional and users can opt-in by checking under settings. The preferences are set in two steps, first, the user details to ChatGPT what they would like the chatbot to know about them. This would be their profession, age, location, programming language, or anything that their responses should reflect.

Secondly, users explain to ChatGPT how they would like for it to respond. For example, for a third-grade science teacher, OpenAI used, “When discussing potential solutions for work-related items, present the information in a table format, outlining the pros and cons of each option—allowing for easier comparison and decision-making.”

Users can also ask for the most efficient solutions, for responses to meet a specific word count or a specific format. The preferences have been capped at 1500 characters and they will be taken into consideration once the user starts a new session.

However, some users feel that this character cap limits the potential of the feature.

1500 character limit though, it definetly feels like an aloha release. For this feature to be really useful it needs a bigger context window. Great step in the right direction though.

— Jairo Rodriguez (@jairodri) July 21, 2023

The feature is currently available in beta to users who have subscribed to the ChatGPT Plus plan but will be made available to other users eventually. The company is yet to avail it to the UK and countries in the EU.

Presently, ChatGPT cannot distinguish where to apply the preferences. For instance, in a test conducted by the Verge, the user input in her custom instructions that she was a coder who worked in the language Golang. When she asked for a chocolate chip cookie recipe, which has nothing to do with coding or Golang, ChatGPT generated the recipe in code.

Therefore, according to the company, the feature is still being tweaked to be able to apply the preferences to the appropriate responses before it is made publicly available to all users.

Inevitably, some users would attempt to use the new feature to get ChatGPT to give responses that would be termed as hurtful or hateful or that violate the company’s guidelines. To curb this, OpenAI is using a moderation API to scan the instructions provided to ensure they meet the guidelines.

Additionally, ChatGPT has been set up to choose not to save such instructions or ignore them altogether when responding.

With providing personal data, as is the case with preferences, there is always the question of data privacy. OpenAI clarified in its announcement that “We may use your custom instructions to improve model performance for our users.” However, users have the option of opting out by disabling the setting in data control.

Related Articles:

• 50+ ChatGPT Statistics on Usage & Revenue for July 2023
• ChatGPT Checker Tools: Revealing AI Essay Detection Loopholes
• MIT Study Breaks Down How ChatGPT Can Speed Up and Improve Simple Writing Tasks