I've worked with technology and business transformation for years, but AI has brought a whole new set of tools to master.
I wanted to understand AI beyond the buzzwords. Not to become an engineer, but to better understand the technology, its business applications, and the decisions involved in adopting it.
So I decided to learn about AI, using AI.
I had a plan: a structured curriculum, practical examples, and questions that would make me think rather than simply memorize concepts.
Starting With a Plan
I started with:
"Build an AI learning plan around my background and goals. Teach one topic at a time, with practical examples and quizzes."
The curriculum covered AI fundamentals, RAG, model selection, enterprise architecture, and governance.
What I liked was being able to adjust the pace. If something made sense, we moved on. If not, I could ask for another explanation.
"Can you explain that using a fictional business scenario?"
Take RAG (retrieval augmented generation), for example. In simple terms, it allows AI to retrieve relevant information to help generate an answer.
Having worked with enterprise systems and support processes, I could relate to a scenario where employees use AI to find information across an internal knowledge base.
Initially, I wanted to understand how RAG worked. As I became more comfortable with the concept, my questions shifted. What about access permissions? Outdated information? How would we know the answers were reliable?
I was moving from understanding the technology to thinking about the decisions involved in using it.
That was the real value of personalization for me. I could build on what I already knew rather than starting every discussion from scratch.
Learning Through Questions
I had included Socratic questioning in my learning approach. I wanted AI to challenge my reasoning, not just explain concepts.
"Quiz me with realistic scenarios. Make me think through the options."
Some scenarios were surprisingly straightforward. Clear requirements, aligned stakeholders, obvious choices.
If only real projects worked that way!
"What if requirements are unclear or priorities conflict?"
AI generally adapted, making the scenarios more realistic. Sometimes I also realized I needed to provide better context.
As the lessons progressed, I found myself spending less time asking whether an answer was correct and more time considering why one approach might be better than another.
The questions became more useful as I pushed beyond definitions and into tradeoffs.
My prompts evolved too. The original plan remained, but I became more specific about the complexity, depth, and kind of feedback I wanted.
Staying on Track
One question often led to another. Model selection could lead to infrastructure, cost, and governance.
Interesting? Absolutely. But as a program manager, my goal wasn't to become an expert in every technical detail. I wanted enough understanding to ask informed questions, recognize dependencies, and work effectively with technical teams.
"Are we still following the learning plan, or should we save this for later?"
I enjoy exploring new ideas, but I also know how easy it is to go down a rabbit hole. Knowing when to dig deeper and when to return to the bigger picture became part of the learning.
There was also the question of accuracy. AI was teaching, quizzing, and evaluating me.
"What assumptions are you making? Where can I verify this?"
A convincing explanation isn't necessarily a complete or accurate one. Independent verification still matters, especially when learning something that could influence business decisions.
Want to Try It?
This is a version of the prompt I used to guide my learning. It evolved along the way, and I imagine yours might look a little different depending on what you want to learn, your background, and how you prefer to learn.
I want to learn [subject]. My background is [experience], and my goal is [learning objective].
Create a structured learning plan and teach one topic at a time. Adapt to my pace and use fictional examples relevant to my professional background.
Use Socratic questioning, quizzes, and realistic scenarios to test my understanding and reasoning. Challenge my assumptions and include practical tradeoffs.
Let me explore related topics, but help me stay focused on the learning goal. Highlight uncertainty and suggest reliable sources for verification.
So, Can AI Teach AI?
For me, yes, with some important caveats.
I started with a clear goal and a learning plan. What changed was the depth of the conversations. As I understood more, I could ask better questions, challenge the scenarios, and think more carefully about the decisions behind the technology.
AI made it easier to learn at my own pace. But deciding what to explore, what to question, and what to verify remained up to me.
Have you used AI as a learning companion? What has worked well for you, and what have you found challenging?