How To Leverage Your New AI Intern

I recently delivered a one-hour workshop for nonprofit leaders in my local area on “How To Leverage AI to Write Grants and So Much More” - it went sooo well! It was an "exponential experiential learning journey" where I watched the courage to try AI grow in the audience while they learned my playbook of AI Intern engagement tactics and tips, all while watching me execute tasks with my AI Intern in their context. The first feedback I received simply said, "This was kind of mindblowing!" - which I put in the win column! Here is a bit of a play-by-play of what we covered, the first of what I hope are many more opportunities to help people leverage their AI Intern.

Attendee review: "As someone who supports staff with a range of digital literacy skills, I really appreciated Tim's approach to the topic of AI. There was just enough technical information to understand how the tools work, and plenty of practical applications to help brainstorm ideas of how it might be useful in a work setting."

Attendee review: "This blew my mind, the possibilities to support and resource the nonprofit sector are unlimited when we learn how to use this tool well!"

Anchoring in How AI Works

I began with my “large language models as word probability mashup artists” riff - paying homage to the highly accessible and visual Financial Times piece "Generative AI exists because of the transformer " which was pivotal in my understanding of large language models, even with such a the technical title. I continued with a new analogy I had developed using “Mary had a little __” as the base, and then looking examples of different learning sets and the resulting probability of the next word - with a bit of my humor creeping in to the final two ;) well, at least I laughed.

AI as an Intern, not an Oracle

Then I did a segment on “AI as an Intern, not an Oracle” – it’s something you engage with to accomplish a task, not something you go to for an answer. Answers are found at the source. AI mashes up words to create probable and sensical outputs. I introduced my mental model of the canvas, the palette, and the strokes as the way to think about engaging your AI Intern.

Canvas (Output): Define the output that you are going to create with your AI intern. We are going to write a proposal. We are going to build an outline. We are going to conduct and compile research. We are going to create an impressionist painting. You don’t need a ton of detail (although the more the better), and it can evolve along the way, but alignment on the endgame creates the guardrails for collaboration and sets critical boundaries for your AI Intern.

Palette/Paint (Context): Provide additional enriching details - or encourage some research - to provide more specific paint to palette. Each layer of context you provide gives your AI Intern additional colors and hues to use when creating with you. We are going to create an impressionist painting of a forlorn looking 44-year old man imaging a company event he cannot attend. We are creating a proposal, for a nonprofit in McCall, Idaho, called CIMBA at the website (https://www.cimbarides.org ). We are going to identify five foundations that align and are relevant to our mission.

Strokes (Prompts) – It takes more than one paint stroke to create a picture, so be ready for laps of prompts - not a single prompt - if you have a high fidelity output you are trying to create. Working with your AI Intern is a process of PromptS Engineering not Prompt Engineering… And, in the next section I would share the common tasks (strokes) that the AI Intern can excel at - with proper direction.

AI Intern Tasks

Next, we shifted to the tasks I use my AI Intern for that I thought would be helpful for an audience of nonprofit leaders. I have five tasks in my playbook that I introduced: text flex, research, brainstorm, write, illustrate.

In the session, I built out a number of contextual examples, as well as some humorous and personal examples, to demonstrate the task and it's flexibility.

Exponential Experiential Learning

This is when it kinda got crazy… The night before I came up with a strategy for demonstrating each of the tasks in a very experiential way - by figuratively pulling one of the audience member up on stage, sort of like on those old clairvoyant TV shows. The notion of watching someone better than you do a task you want to get better at was partially inspired by my children playing Fortnite.

We would discuss the AI Task (e.g. text flex), then look at some generic but relatable examples, and then hop right into ChatGPT to execute the task in real-time for the nonprofit that was our case study. The audience watched as we (me and my AI Intern) built out a series of deliverable before their eyes using the task they just learned. Then, back to learn about the next task (e.g. "research") and back to ChatGPT to watch the next deliverable be built for the nonprofit case study using that task. Jaws were literally dropping, and they learned how to do some prompting!

But, what struck me was that they weren't just learning some great tips on prompting; I actually watched the fear of AI dissipate from their faces.

They got to watch me do it, so it was safe. They picked up on how I was phrasing my prompts. And, it was incredibly personalized and contextual.

All of My Live Demo Prompts

Here is the series of prompts I used to demonstrate the first four AI Intern tasks that would be useful to them: text flex, research, brainstorm, write. For this example, I use the nonprofit Idaho Partners for Good , led by Blossom Pua Johnston , who graciously hosted the workshop.

Text Flex – “go read the nonprofit’s about us page – [insert url] – and then write a 150 word version that appeals to local millennial leaders, and a 200 word version that would appeal to local business leaders”

Research – “now go research five national and five local funding sources – as well as corporate and philanthropic opportunities – that would be relevant and aligned” (I actually did these as three separate "research" prompts, and added a "nice job, now put those in a summarized list for me" just for the awe value)

Brainstorm – “very nice. now do some brainstorming about possible collaboration partners in the area - conduct research and identify three specific potential partners and three collaboration ideas with each”

Write ?– “good work. let's build out partner 2, idea 1. please review that idea, and both the organization and the partner, and write a compelling proposal for it”

After the initial "write" output, I demonstrated "write/expand" and "write/text flex" - a few extension tasks I thought they might find useful. And concluded with the always-sure-to-amaze "illustrate" for an idea from the case study.

Write/Expand – “that's really nice. can you please expand on the objectives section, adding rich language, real data points with linked external references, and a strong connection to the value to the community”

Write/Text Flex – “nice work. now, as a marketing expert, can you write a concise, exciting launch announcement for the effort that could be sent to local media outlets”

Write/Text Flex – “very nice. as a social media expert, write a tweet about the launch, three ideas for Instagram posts, and a Facebook post”

?

Illustrate - "great job. create a photorealistic image to accompany the first instagram idea"

The outputs we got were quite good – we were defining very specific canvases and adding very rich paint to the palette. But what I also noticed was the increasing confidence of those watching the session.

“AI” and “I don’t know what to prompt” began to fade away, as the focus became “how do I define the canvas” and “how can I add rich paint to the palette” to leverage my AI Intern - all in a safe environment, watching someone else do the driving.

I had a number of examples built for us to review - where I would show the canvas, the palette, and then the specific strokes/prompts that I used (tagged), and the output. These were pretty robust examples for:

• Bio Text Flex - translating my full 300 word bio into different lengths, with different tones, and intended for different audiences
• Team Formation and Invitation - brainstorming groups throughout the community that could unite at the intersection for the greater good
• Case for Support - building an outline and then writing each section of a sophisticated fund-raising document
• Grant Synopsis, Outline, Response - creating submitter synopses, provider provides, and documents of alignment and opportunity
• “Write Like Me” - creating a writing style guide template, analyzing a set of my own writing, and producing a "Tim's Writing Style Guide" that could be used to create more content in my voice

Conclusion, For Now, I Hope

While we didn't get a chance to explore all of these examples, I don't mind one bit. To watch the audience learn and experience the prompts in such a hyper-personalized, real-time context was magic. These leaders now have an AI Intern that they can tap into with greater confidence and capability. What an enriching learning experience, for me and hopefully for them too!

I hope to replicate it many more times.

#####

Tim Creasey , Chief Innovation Officer at Prosci , is a researcher and thought leader on managing the people side of change. His work forms the foundation of the largest body of knowledge in the world on change management. He has spoken to hundreds of thousands of people across the globe, helping them to see how successful change is unlockable when we prepare, equip, and support people through change.