I have something I need to share. I’ve been cheating on my brain. I’ve probably written more than 30 million words in my life (total S.W.A.G.) - writing emails, proposals, project plans, policies, procedures, presentations, marketing content, workbooks, training material, etc. etc. etc. I’ve done my share of Google and Wikipedia research, paraphrased from books (with and without source references), and leveraged a serious amount of industry standards and regulations in my writing as well. But at this very moment - I am back with my brain 100%.
So recently things have changed. I began messing around with ChatGPT 3.5 in mid-January, spending a lot more time with “it” since then, trying out all sorts of exercises with outlining concepts, list creation, content support, methodology summaries, job descriptions, child-psychology recommendations, and the list goes on. To say the least, it has been mind-bending.
I actually had a session where I experienced chills, which has happened few more times since, but after the chill session I decided to step away and go on a hike-run in nature to decompress. During the hike I called my daughter, a freshman Forensic Science major at ASU, and asked her what she knew about ChatGPT. She said she had only heard of it, and I proceeded to spew at her my perception of how the proverbial “game has changed”. "Everything is going to be different very soon". "It’s the next coming of electricity, the telephone, the Internet", blah b-blah blah. Her response was “okay dad - chill”. Fast forward three months and she is now coming up to speed - Se la vie.
If you haven’t at least explored it, you must! Go to https://openai.com/blog/chatgpt, sign up, watch some YouTube videos, or call me (916) 500-9726 and I’ll walk you through it. Try starting with this - “Help me write a letter to my wife (or husband) explaining to her how I have been experimenting with ChatGPT and I think I have a new best friend that is going to take up a lot of my time”. I just ran it and the result was hilarious, thoughtful, and in a bizarre way - politically correct. Write a poem, write a thank you letter, write an outline for a master service agreement for your specific industry (be as specific as you want). If you have already been exploring - you already get all of this. Enjoy!
Before I finish this up, let me explain another chill-experiencing moment. Last week, I got an idea to see what would happen if I went down a ChatGPT rabbit-hole with an attempt to create a full set of procedures templates for the “average business entity”. Maybe I shouldn’t even use the word “templates” because I went as far as to have ChatGPT write the actual procedures using “suggested best-practices”. Here was the basic dialog I used:
“As a business owner, help me write a standard list of business procedures, separated by departments that you find within most business entities”
The response gave me seven (7) departments with nine (9) procedure sets each. I say “sets” because they were extremely broad (i.e. “Monitor for security threats”, “Manage social media accounts”, etc.). But it gave me 63 solid procedure sets
Then I continued the dialog with “Now, for each of the procedures in the 7 departments, help me write a list of sub-procedures for each primary procedure.
It killed it. It gave me 9 (not sure why it kept going to 9, but you could also ask it for 20 by being more specific), but it gave me 9 for each of the 63 sets. So now I had 567 procedure titles across 9 groups, in 7 departments each. I read parts of the list and they were solid.
Now, knowing that there is a limit of around 3000 words between the question and response dialog, I decided to begin breaking things down. I copied 1 set of 9 procedures and asked: “Using the following set of 9 procedures, help me write real procedures for each of the 9. For each procedure, use the following descriptive elements: Procedure Number, Procedure Date, Title, Department, Owner, Participants, Description, Inputs, Outputs, Process, Safety Concerns, Security Concerns, and Cross-Department Considerations. Here are the 9 procedures: (and I copied and pasted the department, procedure set titles and the 9 procedures)
The results were relatively generic, and in some cases suggested how to go about identifying the proper response, but overall it was impressive. At a minimum, it just completed 75% if the work that would have taken a human a 1-3 days to complete (i.e. write 9 procedures)
I then asked it to rewrite the 9 procedures using suggested best-practices for procedure specifics that the average business entity would carry out.
It did improve in most cases. We at least had solid procedure models to start from,
So I did all of that work in about 30 minutes, started over only once, and if I had carried out writing all 63 sets of 9 procedures, and put them into 9 binders (or drives) the efficiency math behind it is jaw-dropping. Most organizations don’t even have that depth of procedures let alone the procedures all structured in the same way (yes, they do need to be copied and formatted). The real work left, tailoring them for each department, could be done in a few weeks to a few months vs. potentially years - or never in a lot of cases.
I think the message here is pretty clear: AI is for real, and it’s only going to improve and become more powerful. If I had been more specific about the industry, supporting technologies, company size, size and structure of departments, etc., I am confident it would have provided much better results. At a minimum, and I don’t care who says what in the media about AI being (or not being) an economy and workforce game changer, I personally believe it is. That said, I am firmly in the camp that we need to work with it and figure out how it can make our jobs and lives better vs. being worried at any level. The bus has left the station and there are zero ways of stopping it.
If anyone wants a brief tutorial - call me and we can Zoom one up free of charge. Or, if you would like to request a customized training session in your organization, I would be happy to provide that service. Carpe Diem - ML
Commentaires