GOOD AI prompters know HOW to ask questions. GREAT prompters know WHAT questions to ask.
Reminder: The third Augmented Intelligence live masterclass is today at 12pm EST. The Zoom and calendar links are at the bottom of this email, along with links to the recordings of the first two classes.
Over the last 18 months, I’ve asked chatbots thousands of questions.
Through all of the noise, I’ve noticed four things that are fundamental to success at prompting…
The skill of AI prompting is based on asking good questions
There are two ways to get better at asking questions
One way is significantly higher leverage than the other, yet is overlooked
Mental models help you ask great questions that others can’t even fathom
#1. The skill of AI prompting is based on asking good questions
The metaphor I use for prompting is that AI is like a genie:
Except, rather than rubbing a lamp and making wishes, you type in text and ask questions.
And instead of being restricted to three questions, the limit is how much time you have to ask them. Time is the limiter because there are so many questions to choose from.
So the fundamental question is:
How do you improve at asking questions?
#2. There are two ways to ask better questions
In my experience there are two ways to ask better questions of AI:
Know WHAT to ask
Know HOW to ask for it
You learn HOW to ask better questions by:
Asking lots of questions and noticing what AI can and can’t answer well.
Learning from other people’s best practices.
Improving the quality of your questions accordingly.
For example, some hacks for HOW to ask questions include:
Asking AI to play a role
Providing examples
Being specific
Explaining who the user is
Explaining your goal
Suggesting data sources to pull from
99% of how-to prompt better content on the web is about these types of hacks.
While these are essential, they are the easiest to copy. Therefore, they will only take you so far relative to other people.
By far, the biggest opportunity is knowing WHAT to ask…
#3. Knowing what to ask AI is where the most opportunity is
Knowing what to ask AI is more important than how to ask it, just as a world-changing idea with good enough writing is more valuable than a mediocre idea with perfect craft.
Furthermore, copying other people’s questions will help you get similar results as them.
Learning to develop your own unique questions will help you get extraordinary results that others can’t match.
So, the question becomes, how do you determine what questions to ask…
#4. Mental models help you ask great questions that others can’t even fathom
The reason knowing what to ask is hard is that it depends on your past knowledge and mental models.
Let me give you an example.
I focus on using AI for augmentation rather than automation. The reason I have the augmentation vs automation distinction is because I learned about the mental model of cognitive artifacts.
From there, I focused a lot of my efforts on augmented learning, because:
I think studying how to learn better and faster is one of the highest leverage things to learn in life.
I have built up detailed models of how learning happens over the years.
For example, in Research Reveals The №1 Life Skill That Schools Surprisingly Don’t Teach, I share the universal and fundamental mental model that all learning systems (humans, animals, AI) share:
Because I’ve explored many different approaches to finding the most valuable information in the world, I’ve settled on breakthrough knowledge as the highest-leverage knowledge to learn long-term. Breakthrough knowledge is novel, life-changing knowledge that exposes you to the limits of your current paradigm of living and convincingly explains the benefits of a better paradigm.
How did I settle on the power of breakthrough knowledge? More models…
Because I’ve studied the leverage points model of famous systems theorist Donella Meadows, I understand that a paradigm shift is one of the highest-leverage ways to alter a system.
Because I’ve studied the scientific method, I understand the power of disconfirming evidence. One piece of disconfirming evidence can be more valuable than a million pieces of confirming evidence. For example, every day we watch the sun circle the earth with our own two eyes. Yet, no quantity of observations makes it true.
To summarize, my mental models of cognitive artifacts, learning, breakthrough knowledge, leverage points, paradigm shifts, and the scientific method were fundamental to helping me narrow down what questions to ask.
Did you notice a pattern to what helped me decide what to ask?
Mental models.
Mental models reveal the elements, cause-effect relationships, and dynamics of how things happen in the world.
And, if I didn’t understand the dynamics of learning at a deep level, my questions to AI bots would be general and mediocre, and I wouldn’t be able to evaluate the quality of AI’s responses.
Over the last 5 years, I’ve spent thousands of hours creating 50+ 10,000-word mastery manuals and classes in the Mental Model Club. I had no idea that perhaps the largest benefit of creating these manuals and classes would be that I could use them to 10x the value I and future students would get from AI.
Today’s Agenda: Augmented Mental Models
Today, at 12pm EST, I will share the following with you:
The top 10 mental models I recommend learning to help you with AI.
How to identify the top mental models relevant to your field.
How to use mental models to ask better questions of AI.
And lots more…