May 30 (edited) in Prompts
Intro to Prompting and Chat-GPT
Prompting is the process of having a chat with an AI (Artificial Intelligence) or ML (Machine Learning) model using natural language. When you ask GPT a question or give it a request, it takes what you say and responds in the same language you used. This process, known as NLP (natural language processing), lets the AI understand and generate human-like text based on your prompt.
--------------------------------------------------------------------------------------------------------------------------------------------------------------------
What Does a Prompt Consist Of?
Prompts can vary depending on what you need, but they generally have four main parts:
- Instruction: A specific task or instruction you want the model to perform.
- Context: Any extra information or background that helps the model give better responses.
- Input Data: The question or input you need a response for.
- Output Indicator: The type or format of the answer you're looking for.
Not every prompt will include all these parts, but knowing them can help you create better prompts for different situations.
--------------------------------------------------------------------------------------------------------------------------------------------------------------------
Practical Uses of Prompting:
Prompting is super useful for everyday tasks, especially for research and writing help. But "research" here is a bit broader than the usual digging through books and articles.
Let’s look at an example of how research with GPT can work
I personally like to hike and fish and I use GPT to help me find new places to visit all the time, as well as planning weekend trips.
So let’s ask GPT “What are the best options for late spring hiking within 40 minutes of [my location]?”
(Photo 1)
This will give you a very neat list of hiking trails and parks near you that you can visit, you can also further research individual parks by simply asking questions about some that have been listed.
“Which of these parks has the longest trail loop or distance?”
(Photo 2)
A nice concise answer to help us narrow down which park we might want to visit.
-------------------------------------------------------------------------------------------------------------------------------------------------------------
Let’s do a more complex example. In this case we’re going to assume you’re trying to write a presentation or paper on the medicinal properties of puffer fish venom (I made this up and have no idea if puffer fish venom actually has medicinal properties) and you need some sources of information but don’t want to spend the time scouring the internet and research papers to find quality papers and data.
(Photo 3)
Now let’s ask for some supporting scientific papers or data that support this:
(Photo 4)
It searched the internet and gave us some papers to read on the subject in a matter of seconds. I can’t guarantee that these will all be citable but there’s a good chance they are. This alone would have just possibly saved me hours of research time.
-------------------------------------------------------------------------------------------------------------------------------------------------------------
Now let’s see how Chat-GPT can improve our writing in almost any scenario.
Scenario: We are trying to write a caption for a post on instagram and we can’t seem to find a good balance of witty/funny/applicable vibes.
Lets use this image:
(Photo 5)
So we can actually just copy paste this photo or upload it directly into Chat-GPT and ask for a caption.
(Photo 6)
And as you can see, it actually did a fairly good job. It’s a bit cringe or generic so we can do some further prompting to get closer to the desired outcome.
(Photo 7)
Now this actually isn’t too bad. Giving GPT a specific voice or wiring style to model is usually one of the easiest ways to get the desired outcome more quickly.
But many people will give a caption or sentence to Chat-GPT and just say something like:
“Can you improve this sentence?” Or “make this sentence better”
Typically doesn’t work well because it’s too general and ambiguous. Better or improve is subjective and can mean anything in this context.
When prompting you always want to be as specific as possible, and provide context.
Let’s try another way to improve this caption that I’ve recently discovered myself:
“Can you rate this [sentence, caption, thought, poem or anything else] 1-10 on a scale of [whatever result you’re trying to achieve]?”
For example: “Can you rate this instagram caption 1-10 on a scale of engage-ability?”
(Photo 8)
It’s already given us some instruction on how we could improve it, and also gave us newly revised version based on the improvements. But lets take it a step further and get concrete explanations, so we’re not only improving the caption but learning how to make our writing more engaging simultaneously.
“What are some steps we can take to make this a 10, and can you explain the reasoning behind them?”
(Photo 9)
I didn’t capture the entire response, but it was fairly detailed.
These are just a few ways to start saving time with Chat-GPT right away and get your feet wet with prompting. Remember, context is king while prompting.
We’ll dive into more advanced prompting techniques in laters posts, such as: Output specific, CoT (Chain of Thought), ReAct, RAG, and much much more. And we will also dive into some other features of Chat-GPT like: Code assistant, Dall-E, and audio & image recognition in later posts.
What we covered here today barely scratched the surface.
Thanks for reading if you made it this far, and I hope you found it useful :)
Stay tuned for more tips and examples in future posts!
-Kai
1
0 comments
Intro to Prompting and Chat-GPT
skool.com/elite-wantreprenuers-6277
Making AI and AI tools accessible to anyone, regardless of their technical abilities.
powered by