No Content? No Problem!
Earlier this week, I was on a panel of experts speaking with leaders in education about career opportunities and job prospects in our respective industries. It was a big panel. My role was to represent the software engineering career path and talk with school leaders about the long-term job prospects and career mobility outcomes for these roles.
When I attended a similar event last year, I prepared a set of slides in advance, drawing upon some of the common research on salary bands, job opportunities for software developers in tech and in NYC, and a few slides where I pulled in entry level job descriptions from software engineers.
This year I didn’t have time to update slides in advance. So instead I made a custom GPT to help me prepare instead. It took me about 25 minutes. Here’s how I did it.
![](https://paragraph.xyz/editor/callout/information-icon.png)
First, I created a prompt around the idea that I wanted the GPT to act as a respected educational researcher in New York City to help me create frameworks and compelling arguments for technology access, software engineering career pathways, and project-based learning.
Then I uploaded my slides from last year as foundational content to provide background context in the model. While still in prompt builder mode, I used a separate instance of ChatGPT to quickly generate 30 different statistics that I thought the school leaders might ask me about. I asked for stats on job outlook for software engineers in 2024, both in NYC and nationwide. I also asked for stats and anecdotes about the effectiveness of project-based learning curriculum as a teaching practice. And I asked for salary band details for software engineers in New York City.
Then I pulled all of those notes into a document and uploaded that as additional background content for my GPT.
When I was finished, I typed in three successive prompts:
Hi, I'm a principal at a public high school in the South Bronx. Can you help me make a case about the importance of project-based learning to teachers who are skeptical about why we need to teach kids how to code?
How would I start the process of creating a software engineering pathway at my school? What kinds of projects could we do?
What are the 1-3 first steps I would need to take to move this initiative forward?
I read through the answers and memorized a few labor market stats. I also noted the structure of how the responses were framed to me. Because I’d asked the GPT to essentially put on their “academic hat” when responding, I liked to see the structured way that they framed arguments using data in each point. This gave me some important insights into how a real life group of school leaders might best receive new information about programming and curriculum building.
I borrowed the high-level approach to that structure – include a piece of data, include an anecdote, include a bite-sized, recommended framework or actionable takeaway – in all of my responses on the panel. Based on reactions in the room at the time, this seemed to resonate.
Is this cheating? Or is it simply the new way to work?
![](https://storage.googleapis.com/papyrus_images/562b389c3a6800aa7befbe0bc799f5e0.webp)
How AI Became My Essential Workflow Tool
When I worked at Stack Overflow, we used to joke internally that if our website was down, software developers couldn’t get any work done. We knew this was true because on the (rare) occasions when Stack Overflow was down, developers all over the world would take to Twitter and complain about it. We liked these funny anecdotes so much that we used screenshots of these tweets in our company pitch decks to demonstrate how much people relied on us all over the world.
![](https://storage.googleapis.com/papyrus_images/5aac7617d960f399fe554258a22a5794.png)
There was a lot of truth to this. What we discovered, essentially, was that using Stack Overflow had become a new habit in the workflow of a software developer. Whereas the web 1.0 version of a developer getting answers to their programming questions was, “Read the manual,” the web 2.0 version was, “copying and pasting from Stack Overflow.”
Why am I sharing all of this with you? Because I think the stage we are at with AI right now is incredibly similar. Today, I can’t get any work done without ChatGPT or some other AI tool constantly open on my desktop. When ChatGPT goes down, I’m the one taking to Twitter to see who else is feeling the pain.
This is how I know I’ve built AI as a habit in my workflow. It’s taken the place of what used to be a much more time-consuming task for me in basically all of my disparate job functions: Creating first drafts.
It’s not hyperbole to claim I am at least 30% more productive at work due to my prolific use of AI to help me move projects forward. To me, it feels like using an e-Bike vs. a regular bike. I’m still pedaling; I just get an energy boost so that I can go further, faster.
Reclaiming Time and Creativity with AI in Education
Later on in the session, one of the school leaders asked me in a smaller group session about how important it was for them to teach their students effective use of AI.
I shared the story of how I’d prepared for my panel and my workshop with all of them. I pulled up ChatGPT and glanced through the stats and research it shared with me in real time. I told them how it’s largely thanks to AI that I’ve been able to take on work projects that would have previously seemed unfathomable or impossible to wrap my head around as a solo operator. I reminded them that somehow I had figured out how to use the skills acquired from an academically trained journalism degree and apply these in a general management context which is how I had even a fighting shot at staying afloat in deeply technical roles I threw myself into in the crypto industry.
I also relayed how much fun it is to learn something new. I’m talking, good-old-fashioned, what’s possible here, let’s figure it out together, fun. Presumably the kind of fun that any school leader or teacher felt unlocked for them when they first decided to wear the badge of honor and commit their careers in service of lighting up learning light bulb moments for thousands of others.
I acknowledged that it’s hard to find fun in education today with all the accountability, testing, checklists, frameworks, standards, pandemics, racial equity issues, curriculum design, student crises, TikTok threat campaigns, paperwork, supply shortages, budget cuts, and fighting over copy machines. I recognize that I have more time than classroom teachers who show up for kids every day. To me, time is a luxury.
But part of what gets me excited is that, when AI becomes an embedded habit in all of our workflows, we might get a little of that time back again. Time to feel creative. Time to remember why we love to learn. Time to experiment and tinker and feel the little dopamine surge of creative ideation that you can only get when learning something new. (By the way, this feels a lot better than the dopamine kick when someone likes one of your social media posts.)
So, do I think schools need to teach AI to kids? Absolutely. But the reality is, many students are already exploring AI on their own. Maybe instead, we might consider: How can AI support an over-tasked collective of school leaders around the world who haven’t had a chance to catch a break since the pandemic?
By embedding AI in our workflows, we can all reclaim a little time, enhance creativity, and foster a learning environment that thrives on innovation and curiosity.
![](https://storage.googleapis.com/papyrus_images/d31c06b5234a02946765656e3fdd1974.webp)
![highlight](/api/highlight?text=Beyond%20the%20Classroom%3A%20Leveraging%20AI%20for%20Educational%20Innovation&author=Hard%20Mode%20First&url=%40bethanycrystal&avatarUrl=https%3A%2F%2Fstorage.googleapis.com%2Fpapyrus_images%2F4f830df62c48bee65c13da9924e4ad76.jpg&featuredImageUrl=https%3A%2F%2Fstorage.googleapis.com%2Fpapyrus_images%2F690202c3dbaf0bdb895c0bb7863478ce.jpg&size=512)
![Hard Mode First logo](/_next/image?url=https%3A%2F%2Fstorage.googleapis.com%2Fpapyrus_images%2F4f830df62c48bee65c13da9924e4ad76.jpg&w=384&q=75)