100 Days of AI Day 4: Maximizing Productivity & Creativity with ChatGPT

I write a newsletter called Above Average, where I talk about the second-order insights behind everything that is happening in big tech. If you are in tech and don’t want to be average, subscribe to it.

LLMs are trained on the open internet. And open internet has knowledge about all sorts of topics which LLMs acquire. They are not 1–% accurate, but they are useful without a question.

LLMs are really good at manipulating the text knowledge. Like most of you, I often use chat-got to get started on emails, write an invitation RSVP I am about to send out, reword my weekly newsletter, and do other mundane tasks that I might need some help with. In these cases, ChatGPT acts as a creative partner and sort of replaces your friend or colleague that you would otherwise use to brainstorm with (Let’s not get into the will we replace our human connections part of the discussion this early on Day 4 of 100 days of AI).

But the point is clear: using chat-gpt, you can be more productive and creative in your daily job. So, knowing some techniques below will help you keep above the crowd.

Leverage Transforming Capabilities:

LLMs can recognize which language a text is & translate text from one language to another. They can act as a universal translator. Let’s say you are reviewing customer feedback for you product from Japanese customer, you can ask Chat-GPT to translate into English to quickly go through them. I would suggest having an IDE setup with your basic code always ready to be more productive than interacting with LLMs with the default UI interface they provide. In the case of Open AI, you have chat-gpt. But I would suggest using whatever IDE you have to set up a boilerplate query to be more productive. I use Python + VS Code. Here’s my starter code for writing queries.

from openai import OpenAI
client = OpenAI()
client.api_key = 'YOUR SECRET KEY'

text_1 = f"""
Making a cup of tea is easy! First, you need to get some  
water boiling. While that's happening, 
grab a cup and put a tea bag in it. Once the water is
hot enough, just pour it over the tea bag.
Let it sit for a bit so the tea can steep. After a
few minutes, take out the tea bag. If you
like, you can add some sugar or milk to taste.
And that's it! You've got yourself a delicious
cup of tea to enjoy.
"""

prompt = f"""
You will be provided with text delimited by triple quotes. 
If it contains a sequence of instructions, 
re-write those instructions in the following format:


Step 1 - ...
Step 2 - …
…
Step N - …


If the text does not contain a sequence of instructions,
then simply write \"No steps provided.\"


\"\"\"{text_1}\"\"\"

"""

response = client.chat.completions.create(
  model="gpt-4",
  messages=[
    {
      "role": "user",
      "content": prompt
    }
  ],
  temperature=0,
  #max_tokens=64,
  #top_p=1
)

print(response.choices[0].message.content)

Transform The Temperature:

Note that most people do not change temperature while querying chat-gpt. Changing the temperature is a way to control how random (in other ways creative) you want your answer to be from Open AI LLM. In the above code snippet, the temperature is set to 0, but you can set it to any value between 0 and 1.

When you are writing an invite to a party, and your vibe is not about sending a regular message, you should use the temperature value as 1, but if you want the most standard way of writing something, say you want a starting draft for writing an email to your local congressman, then you should be using temperature set to 0. To show how this works, I have asked chat-gpt to write an invite for a board game & BBQ party I am hosting this coming Saturday but used different temperature values. Here’s what it gave me.

Query: Write an invitation for a Board game & BBQ party at my place on Jan 6th. Keep the response to 2 lines.

Response (temperate = 0): You’re invited to a fun-filled Board Game & BBQ Party at my place on January 6th! Enjoy good food, great company, and exciting games. RSVP to confirm your attendance.

Response (temperate = 0.3): You’re invited to a fun-filled Board Game & BBQ Party at my place on January 6th! Enjoy good food, great company, and exciting games. Please RSVP and let the games begin!

Response (temperate = 1): You’re invited to a night of Board Games & BBQ at my place on Jan 6th! Enjoy savoring delicious grilled treats while winning fun games. Kindly RSVP soon so I can save a seat for you!

As the temperature value increases, the responses get less formal and more random/creative. Try this the next time you have a query.

Expanding:

A lot of what we ask chat-gpt to do is to expand. Meaning that we give it a short text and ask it to come up with a longer text following our instructions. And LLMs are great at expanding.

You can use expanding to automate your replies to your customers based on whether the received email is positive feedback or negative. The prompt would look something like this.

Prompt:

• You are an AI customer service agent.
• Your task is to send reply emails to customers.
• Given the customer email delimited by “`,
• Generate a reply to thank the customer for their review.
• If the sentiment is positive or neutral, thank them for
• their review.
• If the sentiment is negative, apologize and suggest that
• they can reach out to customer service.
• Make sure to use specific details from the review.
• Write in a concise and professional tone.
• Sign the email as AI customer agent`.
• Customer review: “`{review}“`Review sentiment: {sentiment}

AI PRODUCT IDEA ALERT 1: Gmail could give users an option on how to respond to emails using AI for all of us. This might require more features along with an ability to tag emails for which we want to respond and what additional steps and even an ability to give a custom prompt. The difficulty might be how would be here to identify what % of this system would fail and how can Gmail give the ability to reduce that for the user.

AI PRODUCT IDEA ALERT 2: A Shopify app for merchants to respond to customer requests & feedback. In this case, the App would generate drafts based on the rules decided by the customer, or the app can itself generate responses appropriately. And create a draft to review for the merchant. This could help solo merchants or merchants with small support teams save a lot of time on support and processing requests.

That’s it for day 4 of 100 Days of AI.

Follow me on Twitter and LinkedIn for the latest updates on 100 days of AI, or bookmark this page.

Previous - Day 3: How to Use Inferring to Create new AI Features?

Also published here.