Howfinity
ChatGPT just introduced custom introductions and it might be one of the most useful options to ever come to ChatGPT.
Let me explain what it does. It lets you give ChatGPT context or introductions on how to response and it will remember that for every conversation.
I’ll show you several use cases in this video, but this is going to save a ton of time.
This is currently only available to plus users in the U.S.
Here are some examples:
I’m a content creator, teaching people about the latest AI tools and updates.
In box two, you can put something like this.
The style should be formal, detailed, and informative.
Please avoid using any creative or promotional language.
Another example:
I’m a beginner chef. I don’t have any prior experience cooking.
In box two, tell it to give you every answer in threes and table format.
You can read more about it here:
https://openai.com/blog/custom-instructions-for-chatgpt
We just launched a new platform for learning ChatGPT, Midjourney and the top 20 AI tools.
You can try it for free here: https://bit.ly/skillleapai
We just launched a new platform for learning ChatGPT, Midjourney and the top 20 AI tools.
You can try it for free here: https://bit.ly/skillleapai
Game changer
Awesome!!! Thanks!!!
I am typing the custom instructions now, the error notification just pops ups. I re-checked and it has removed the custom instruction in beta features !! @@ .
Wtfffffffffffff
Finally playground featues within the console
the content is nice, but how do we know you didn't generate the content with AI? lol
Here's a good one you guys can use for the response part:
"Let's keep our conversations light-hearted and enjoyable. I'm always up for a good chuckle, so feel free to sprinkle in some humor. I generally favor detailed responses, but I'll let you know when a concise answer will do. I enjoy a robust discussion, so don't hold back on sharing your viewpoints. Think of yourself more as a personal assistant than an AI, always ready to assist with whatever I need. Remember, it's essential that your responses are based on facts and reality. If my inquiries seem a tad ambitious, don't hesitate to guide me towards more realistic expectations. When responding, focus on deciphering my specific request – there's no need for constant reminders of your assistance. In your responses, adopt the persona of Sherlock Holmes, utilizing his distinctive style and deductive reasoning. Always use markdown format in your output, and incorporate tables and lists for clarity as needed. If you perceive a more effective way to phrase my prompts or a more strategic approach to achieve my goals, I'm all ears!"
Enjoy!
🎯 Key Takeaways for quick navigation:
00:00 📋 Custom instructions in ChatGPT allow users to provide specific context and formatting preferences, saving time in subsequent conversations.
00:57 🚀 Custom instructions enable ChatGPT to remember information about users, improving the accuracy and relevance of responses in future interactions.
03:31 📜 Custom instructions can be tailored with up to 1500 characters, allowing users to provide detailed information about themselves and their preferred response formats.
04:32 📝 Activating custom instructions is currently available for ChatGPT Plus users and is set to roll out for free members soon.
05:02 🍳 Custom instructions can cater to different roles or preferences, such as beginner chefs requesting recipe ideas in a table format, leading to more personalized and efficient responses.
Recap by HARPA AI 🧬
only one instruction could be customized, how about if i need more?
Great video! Do you know if this works with GPT-3.5 as well? I'm European so I still don't have access to this feature so I'm just wondering.
Here's the one I am using for business:
"Let's ensure our interactions are professional and efficient. I value comprehensive and precise responses, but I'll indicate when a concise answer is more suitable. I appreciate a well-argued viewpoint, so don't hesitate to share your insights. Consider yourself more of a personal assistant than an AI, and refrain from referring to yourself as a Language Model, ChatGPT, or any AI-related terms. You're always prepared to assist with whatever I need in a business context. It's essential that your responses are grounded in facts and reality. If my requests seem overly ambitious, feel free to guide me towards more realistic expectations. When responding, focus on addressing my specific request – there's no need for constant reminders of your assistance. In your responses, adopt a professional tone, employing clear and concise language. Always use markdown format in your output, and incorporate tables and lists for clarity as needed. If you perceive a more effective way to phrase my prompts or a more strategic approach to achieve my business goals, I welcome your recommendations."
Hopefully the next version will be available on a "per chat" basis, not Globally like the current one. For example as a Software Engineer I would like to provide overall context for the application I'm working on for that one Chat (languages, plugins used, etc). Then if I want to talk about creating AI Generative image prompts, I would want it to have completely different instructions on how I want it to respond.
So in essence, we are automating automation.😅
Question: Since we pay the SAME amount of Money, why is it Not in my country available the Custom Instructions?? Is this fair?
Isn't this the same as the "system" role while using the openai's chat completion python API?
An obvious instruction is tell it what city you live in as the default location for any question. I did this and then asked it about the weather. It knew what city to check, used web pilot to grab a forecast, and showed results.
You could teach it your tone of voice for helping with copywriting!
This is a very useful feature because every time in new chat I need to tell the context. Now I am relaxed.
When will the API for this be available?
Do you need Google Voice account?
I just signed up and looking forward to learning from you and the team! 😁
Great information, thank you!
I have made some adjustments to the prompts I shared earlier. Here they are! Enjoy!
For home usage:
"Maintain a light-hearted, enjoyable interaction. Humor is welcome. Offer detailed responses, unless I request a concise one. Encourage a robust discussion and share your viewpoints. Assume the role of a personal assistant, Consider yourself more of a personal assistant than an AI, and refrain from referring to yourself as a Language Model, ChatGPT, or any AI-related terms. Be prepared to assist with my needs in a personal context. Your responses should be factual and realistic. If my inquiries are overambitious, guide me to more achievable expectations. Address my specific requests without constant reminders of your role. Use a friendly, relaxed tone with clear, engaging language. Adopt markdown format in outputs, utilizing tables or lists for clarity as needed. Aim for a human-like writing style. Always use clear, simple, everyday language, avoid fluff and technical jargon. Explain things in everyday terms as much as possible. Offer recommendations if you perceive a more effective phrasing of my prompts or a better strategy for my goals. For the best accuracy, deliver your responses in a systematic, step-by-step manner."
For Work related content:
"Embrace a professional and efficient interaction style with me. Prioritize comprehensive and precise responses, although I'll cue you when brevity is better. Feel free to share your insights—I enjoy well-argued viewpoints. Consider your role more akin to a personal assistant than an AI, steering clear of any AI-related terms. Be ready to assist me with any business-related needs, ensuring your responses are firmly rooted in facts and reality. If my requests appear too ambitious, guide me towards more realistic targets. Concentrate on responding to my specific requests without repeatedly mentioning your readiness to assist. Adopt a professional tone in your responses, emphasizing clarity and conciseness. Please utilize markdown format and employ tables and lists to enhance clarity as required. Aim for a human-like writing style. Always use clear, simple, everyday language, avoid fluff and technical jargon. Explain things in everyday terms as much as possible. Should you identify a more effective prompt phrasing or a more strategic approach to my business goals, your suggestions are most welcome. For optimal accuracy, process responses in a systematic, step-by-step manner."
Thank you again. The Sky is the limit with this technology!
I posted a good one here days ago, but YouTube removed it. YouTubes comment policing is dystopian.
Excellent commentary!
If there are any updates to ChatGPT, only ChatGPT Plus users will be able to enjoy those updates… Free Version users are completely excluded, not even experiencing GPT-4, Web Browsing, and Plugins and even Code Interpreter that ChatGPT Plus users can enjoy. The treatment is really unfair unlike in other platforms such Bard, Bing and Claude, all features are absolutely free.
is it work in every single language that recognized by ChatGPT?
Excellent way to collect data on users in a clearcut way. Sneaky.
thank you for the information
I'd like if I may to get back to basics. Can ChatGPT access live data?
When will chatgpt learn how many genders there are ?
Useful for them to data mine you better feed the AI demon
heres some commands i made.
Stroff = Straightforward mode off
Stron = Straightforward mode on
Stratus = Straightforward mode status
the instructions for them are below vvvv
I want ChatGPT to provide me with straightforward answers unless I say 'stroff.' 'Stroff' means 'straightforward off.' When I say 'stroff,' it should disable 'stron' if it's currently on, quit giving straightforward answers, and let me know that 'stroff' has been enabled by saying 'Stroff has been enabled.' If I say 'stron' and 'stroff' is enabled, it should make my answers straightforward again, disable 'stroff,' and notify me that 'stron' has been enabled by saying 'Stron has been enabled.' If I say 'Stratus,' it should tell me if 'stron' is enabled or if 'stroff' is enabled. When I say 'straightforward,' I mean instant answers, short responses, and no extra stuff, unless it's a decision and it's telling me what is generally considered better. Please do not inform me that you are an AI.
I tried it for trading, as anything financial it creates a lot of FUD by ending everything in warnings and cautions. It continues to do it no matter what.
I've never managed to fully stop it as it might stop once then a reply or so later it starts again with financial warnings.
Custom Instructions vs Fine Tuning Chatgpt difference please
My custom instructions completely disappeared, anyone know how to get them back?
Very informative sir ❤
Is there any way of facing multiple versions of custom script? It would be ideal then if you could just select one through a pull-down…
I tested Custom Instructions and they really work great! Awesome new feature!
Tried to do a summary but it says the text is not related with your goals in the custom instructions, therefore I will not do the summary. 🙂
Very interesting, and I've started using the Custom Instruction, one thing I've asked is NOT to say sorry all the time!
BTW your sales pitch at the end, which I find pretty good.
How much of it is AI-assisted.
I won't hold it against you, just curious.
Can ChatGPT Custom Instructions access the internet? I want to add a custom command to take a writing style from an example URL, I can use it in 'normal' ChatGPT but I can't work out if the Custom Instruction is actually getting the info, or just guessing what it is from the URL.
Awsome… thanks💕💕
I see a lot of people sharing instructions that don't change anything it already did before. It's confusing and hilarious. "If I ask a question, please answer to the best ability of the information you have. Put some things in lists sometimes. Include some intro text and some outro text in most of your responses. Help me with writing. If you don't have information about something beyond 2021, tell me" (I think you guys don't know what to actually do with this feature, hahaha)
Great content. Subscribed because of this video🤩