How to Become a “Power Prompter”
![Workplace Meeting](/sites/default/files/styles/fullscreen_4x3_767x575_1x/public/2024-07/workplace%20meeting.jpg?itok=Ykn0aro-)
We've all heard the term "power user." Someone who has excelled in some type of technology – or maybe they can simply type fast – that gives them a productively edge when leveraging technology. Many of us in the legal profession – at least the older generations – have long given up hope that we can become a "power user."
But with generative AI, you get one more chance to prove the story that you tell about yourself wrong. Not only can you become a power prompter, but you will. You just need to set aside a little time every day for a month or two and play around with a generative AI tool. Do it!
As Microsoft's Jason Barnwell explains so well in my interview with him, context is key in developing the prompt you will ask your tool. And then the second key component is the iterative process, where you further refine your prompt again and again.
It's just the way your mind works – when you think of an idea, you don't go with the first thing you think of. You spend some time noodling, refining it again and again. Either by yourself or by bouncing your ideas off others. That's what you're doing with the AI tool. Bouncing your ideas off of it - and boy oh boy, you're gonna find it's a smart cookie if you're feeding it the proper prompts.
So start with context and reiteration – but then there are a number of ways to go from there in terms of your reiteration strategy.
Tony Wan has a podcast called "Generative AI" and in an episode dated August 14th called "Can AI Generate Original Content?", Tony teaches us three types of "power prompts":
- Role playing – You tell the AI tool to play a role in a particular scenario that you are working with. Implementing role prompting is as simple as instructing the AI to "you are an entrepreneur" or to "you are a forensic accountant."
- Train of thought – You encourage the AI tool to reason in a way similar to how the prompt is written, typically providing a series of specific steps. You want the AI tool to explain the steps it went through in order to get to its final answer – but you need to give explicit guidance about which steps you want to hear about to get the best results.
For example, you can instruct the AI to write a business plan that covers these five specific areas. You list the areas, perhaps specifying the title of the caption for each – and you give the AI a sense of how in-depth you want the business plan to be. You can specify the type of audience that the business plan would have (egs. big money investors; friends & family investors; community leaders; potential employees).
- Self-critique – You can ask the AI tool to assess its own output for potential inaccuracies or areas of improvement. Also known as "judging" or "self-consistency" prompting, it's like asking the AI to double check its own work so that you foster reliability.
For example, you can instruct the AI tool that "you are 101 entrepreneurs, 100 of whom have a 100-word pitch and one who is the judge who evaluates the 100 pitches to pick the best one." This would take a human a long time to come up with 100 pitches and pick the best one – but you can hand over that task to the AI tool, who would not complain or find it tedious to do so. And it can do it much faster.
Tony talks about combining all three of these prompting techniques into a single prompt to create a "super prompt." And then remember this is a reiterative process, where you go back again and again, refining the AI tool's responses by giving it more and more guidance.
Here's a great 26-minute video entitled "The ChatGPT Playbook of EPIC Prompts." Matt Wolfe illustrates the different ways he uses ChatGPT to tackle all sorts of daily tasks. This should help you get ideas on where you can best use AI tools to assist you in your life and how to best design prompts.
As you will see, there is a little bit of an art to prompting. You will get frustrated at first with the results of your prompts most likely. But that's because you haven't mastered the art of prompting. But there isn't much to it if you stay patient and realize there is a little bit of a learning curve. And generative AI will be dramatically changing within the next few years as advances are coming fast and furious. So the art of prompting should keep getting easier – but it will always be a skill you're going to need to develop.
And to be honest, practicing what you ask AI to do for you should help your own reasoning processes. Even if you give up on AI for some reason, the focus and concentration necessary to create good prompts should help refine your own techniques and practices when it comes to how you go about your daily life…
Print and share
Explore more in
Public Chatter
Public Chatter provides practical guidance—and the latest developments—to those grappling with public company securities law and corporate governance issues, through content developed from an in-house perspective. Subscribe 🡢