Two Questions to Unlock Your AI Engagement

How prompting with purpose will transform your next GenAI project. Recently, someone asked me for advice as they got started with prompting in #GenAI. My immediate response was “what have you worked with it to create?” They gave me a blank stare, surprised by the question instead of Yoda-like answers, and replied “Umm, I wasn’t sure what to ask it.” In that moment, it struck me that A) even from the beginning, my relationship with GenAI has always been one of a collaborative partner to co-create something with, and B) working together with AI starts with having a thing to work on…

This article unpacks some of my initial learnings about the critical context for partnering with AI – having some “thing” to partner on. At the end, I share two AI unlock questions – one to ask yourself, and one to ask GenAI the next time you prompt (so be sure to get to the end).

It’s Called “Generative” AI for a Reason

The word “generative” is an important part of the name. It doesn’t “know” everything, even if it has read everything to understand how words are strung together to make meaning (and then built the capability to also string together words to create meaning by predicting the next word); it is not omniscient intelligence. It is also not a lightning-fast lookup tool or retrieval engine that goes and pulls an answer from somewhere; fetch-it intelligence.

The name matters – "generative" means it creates something new with you - "generate" is what it does.

“Generative” AI does just that; it generates anew using the response fodder it culled and with the instructions you have provided it.

Any Reason Will Do

When I engage my “AI Intern” (a frame I started using in January 2024 to describe this new working relationship), I do it with a “generated output” in mind. This was especially true when I was first learning how to engage AI. And while those early ones weren’t mission critical outputs, they were defined work, and I (or the person I was helping) knew what we were setting out to generate.

• 1960s superhero cartoons aligned to top changes on the horizon in the research to jazz up a webinar.
• Age-appropriate, hyper-personalized coloring book pages for children with recurring characters, because it’s hard to find a page with a unicorn playing with its pet rat or skiing down a mountain for a 7 year old.
• A business plan to create a company to produce said hyper-personalized coloring books.
• Adult coloring book pages representing specific places or memories from a family vacation.
• A draft grant proposal, weaving together information about the nonprofit and the grant provider, neither of which I knew anything about but both of which I knew how to have my AI Intern research.
• Vibrant cartoon characters for "The ABCs of Change" - an alphabet book demonstrating the cost of leaving change to chance and the value of taking charge of change, manifesting a dream that started in 2017.
• An “AI for Change Agents” curriculum outline and syllabus.
• A manuscript of a business book, produced from podcast transcripts and keynote decks, about twelve alignment hacks for "getting on the same page faster."
• Artwork in six styles of the Grand Mesa in Western Colorado for a birthday present.

Even the MyGPTs I created when that capability came online had an “output to generate” baked into them, they too have an output or purpose baked right in: laser-engraving optimized image generator for black slate coasters, a coloring page maker for kids (and one for adults , too), a "case for support" generator for non profit leaders, Timbot 2.0 as a conversation bot trained on my writing style and hundreds of pages of podcast transcripts, a change management prompt generator that turns a change task into a well-worded prompt, an "ABCs of Change" generator to extend the book adding richness to each letter you submit, even a translator that I can drop any chunk of text into and it translates into my voice using a “write like me” style guide.

My motivation varied. My familiarity with the subject matter varied. The outputs certainly varied. However, there were a few universals across the arc. First, we knew where we were going from the beginning. And second, I let AI tell me how to help us work better together.

Two AI Unlock Questions

I’m going to provide two questions that will help you unlock AI – one for you to ask yourself, and the second to ask AI at the end of your first prompt. These questions emerged from the many collaborative journeys with a generative AI Intern. But, you can use them literally the next time you prompt to get better outcomes.

Question 1 - as you start, ask yourself: What are we going to build together in this session?

This is the question I have top of mind, front and center, each time I open up a GenAI window (usually ChatGPT, but a lot of Kaiya these days too). It is the question that the “human in the loop” (you) needs to ask and answer at the start of each of your AI engagements as well. The question sets the stage for the collaborative partnership. It also focuses the session on the output to be created. You may refine or redefine the output along the way, but you’ve started from a place of alignment.

If you are stuck, you can always try a frame like one of these: an outline, an article, an agenda, a synopsis, a plan, a comparison, or an evaluation rubric.

Bonus tip: go ahead and tell AI what you’re going to build together from the start! No need to keep it a secret; you wouldn’t keep the output a secret from an intern you were engaging for help, would you?

Question 2 - at the very end of your prompt, ask AI: What questions do you have for me before we begin?

That’s exactly what you type, at the very end of your prompt. The first time I closed out a prompt with “what questions do you have for me before we begin?” I was starting a big undertaking, and I wasn’t confident if I had provided enough details to get off to a good start. So, I went ahead and asked AI. The response back included five specific questions, related to the direction I was heading, but that gave me the chance to really add the contextual details that would help bring what it created closer to what I had in mind.

As it turns out, each and every time you prompt is a chance to give AI more enriching details to generate a better output. And then best way to know what details to add is simply to ask!

The “Putting It All Together” Prompt

“What are we going to build together?” – to identify the output from the onset, and set the tone for collaboration

“What questions do you have for me?” – to let AI help you add the right enriching details, and to start the collaboration immediately

We can combine these two elements together, along with some pleasantries and a few initial bits of contextual detail, and the result is a sort of meta prompt:

Good morning. I’d like to work with you to build/create ___. Here are a few details to start: ___, ___, and ___. What questions do you have for me before we begin?

If you don’t have a “thing” that comes to mind to try working on right away, here are a few suggestions to try (or imagine and try):

• Plan a party for a family member, friend, or occasion
• Plan a 10-day road trip to an areas you want to visit
• Create a set of illustrations to represent your hometowns
• Write two chapters of fan fiction about a favorite character

And there they are. Two questions that set the tone for collaboration and give AI the chance to help you know how to give it what it needs to help you best.

Go ahead and give it a try! And if you are able, please do share the results of your first "What questions do you have for me before we begin?" as it will help each of gain a better understanding of how to engage GenAI.

#AncoraImparo