Basic Applications

01. Introduction

You can feed the chat bot data in a paragraph and ask it to produce a table for you with the data structured
Generate a table containing this information:
- furthermore you can also ask that the table be in Markdown format for ease of copy and paste

can be used to write emails for you
- Be them simple messages or complex ones.
  - Simple: Write an email to my boss saying that I will be out of office today since I am sick.
  - Complex: Write a humorous yet professional email to my boss saying that I will be out of office today since I am sick. Be concise and funny. Include a funny reason:
  - Emphasizing the seriousness: Write a **serious**, professional email to my boss saying that I will be out of office today since I am sick. Be concise:
Can be used to summarize the data contained in an email as well, saving time
- generate a summary of this and a list of action items
- This can then be fed into the AI as part of a prompt for it to write a response.
Can be used to write the dreaded cold email
- This could be a simple message pitching a product to a founder
- Or it could include unstructured data
One of the great uses and helpful elements of LLMs is that they can cut through information clutter quickly.
- and parse unstructured data quickly

Significant reductions of writing time for blog posts
- process should be iterative which looks like the following:
  1. write prompt
  2. see what model outputs
  3. taking pieces out of the output to put in a blog post
  4. repeat until complete
- Good to start with having it provide an outline
  - if the outline was too long, saying this to the AI and asking for a shorter one is a good next step
    - with GPT-3, the outline will need to precede the new prompt of course
Other Tools for Writing Blog Posts
- Jasper.ai
- Copy.ai
Big takeaway is that while time saving, its not fully its not fully automated to use AI to write blog posts.

Feeding a passage to the LLM and asking what a specific term there within means can render helpful results
They can generate practice problems for upcoming tests
- Can give your notes as context in this process
ADDITIONAL IDEAS - LLMs are excellent at providing the user with topic overviews and lesson plans specific to their requests that may be outside the scope of available tutorials or courses.

Code Generation - can write simple sections of code relatively easily.
Code Reformatting - after prompt include three hashmarks (or other symbol) to delineate the boundaries of what it should read as code
- Please add line comments to this Python code and reformat it for legibility:
Debugging - can determine syntax and logic errors in code fed to it.
- Please debug this Python code:
Code Optimization - Optimization of existing code without breaking the logic
- Act like a very senior python developer. Please optimize this script:
Translating Between Languages - very useful means of converting from one to another, like an older language to a more modern one.
Can Handle Multiple Files - just include the filenames with the pasted codeusing a structure like this:
- BEGIN FILE 1: ./foo/bar.py
Simulating a Database Serer - can simulate server environments without needing to experiment on production databases
Simulating a Web Server - for getting responses to HTTP headers for instance
Simulating a Command Line - for when a VM is just too much hardware overhead
- Act as Debian Linux command shell. Please respond to my commands as the terminal would, with as little explanation as possible.

LLMs can help find the perfect emoji for a situation
- include the post/tweet you need emojis for and
  - What are some good emojis to add to this tweet?
  - or by sentence
  - or even What are some good emojis to add to this part of a tweet?
Always prefer your own judgment when choosing emojis, ChatGPT and its ilk can only assist you in making the choice, not automate away all your work (for now...)
This task may be better with ChatGPT as it justifies its responses whereas GPT-3 will not and just outputs a list of emojis.

LLMs can simplify the process of reading and understanding contracts (and legalese in general)
finding dangerous legal language
- What part of this agreement contains dangerous language?
While it can write various legal documents, the less common they are the more caution should be exercised
Use of LLMs here as everywhere is in an assistive capacity

You can prompt the LLM to use different tones and author writing voices to change the output dramatically.
- this includes your own style, which you would include with the prompt of course (few shot prompting)

can help in digestion of previously written material
- famous works do not require inclusion (like Romeo and Juliet)
can summarize or create an outline of text
- 2,500 words max, longer texts need to be broken into text
- can also provide outlines

Zapier allows using GPT-3
- which really expands the potential of its automation
- IFTTT offers in house AI so the utility of this in particular to me is low as I am used to IFTTT, otherwise I would probably wire n8n for this sort of thing.
take away here is that the LLM can interpret and act on emails easily
- BUT THERE ARE PRIVACY CONCERNS WITH SUCH