When a colleague teaching an intro to technology course approached me about her students’ interest in learning about AI and asked if I could teach it, I was excited. I developed a five-session unit centered around business development with an AI co-founder. What I discovered surprised me, and has implications for how we might think about preparing students for an AI-integrated future.
The Startup Approach
I’ve been following Ethan Mollick’s work and was inspired by his book Co-Intelligence, as I’ve described before. A lot of his examples use his business classes, and that got me thinking. What if students could explore AI literacy through creating their own startup ideas with an LLM (a Large Language Model, like ChatGPT) as their thought partner? This approach seemed promising for several reasons:
- It would naturally showcase both AI strengths and limitations. For example, I suspected that creating a business plan would work great, but getting accurate, comprehensive market research might not.
- The business focus aligned perfectly with the course. Rather than teaching AI in isolation, we could connect it to the broader business and technology themes they were already exploring.
- Giving students agency to develop their own business ideas would make the experience more engaging and potentially useful beyond just learning about AI.
My biggest surprise came early. While everyone (ranging from 10th grade to post-graduate) had used AI tools before, none had experience using them as thought partners in an extended dialogue. Their previous interactions had been largely transactional: asking questions, generating content, or analyzing information.
In other words, they were using LLMs primarily as search engines or content generators, missing out on the more powerful collaborative aspects these tools offer. This limited perspective isn’t unique to my students. I suspect lots of us are still in the early stages of understanding how to have meaningful collaborations with AI.
From Instructions to Insights
The unit progressed through five sessions, moving from LLM foundations and system instructions to market analysis, business plan development, and pitch deck creation. It culminated with presentations in front of a panel of “angel investors” (other teachers) who would decide which business ideas they would back with investment, if any.
One misstep I made was assuming that older learners would benefit from figuring out system instructions on their own. I deliberately avoided modeling this process, thinking it would encourage exploration. In reality, this created unnecessary confusion. During the next class, I needed to meet with each student individually to help refine their instructions for effective AI collaboration.
I also noticed that the AI tended to be overly supportive of student ideas, even when some weren’t particularly feasible. In future iterations, I plan to explicitly teach them to use “adversarial prompting” that deliberately asks the AI to critique their ideas from different perspectives, such as skeptical investors or potential competitors.
Despite these challenges, the discussions throughout the unit were wonderful. Students began to see patterns in effective AI collaboration and gradually shifted from seeing LLMs as simple tools to understanding them as complex thought partners with distinct capabilities and limitations.
By our final session, the transformation was evident. Those who had previously used AI only for quick answers were now engaged in extended dialogues about business strategy, market analysis, and product development. Their presentations demonstrated not just business understanding but a newfound ability to collaborate with AI in sophisticated ways.
During our closing discussion, everyone agreed that the experience had been valuable and changed how they thought about AI. One student even shared that the unit had inspired him to pursue business studies more seriously — a fantastic unexpected outcome.
Lessons for My Teaching
This experience reinforced several key principles for my teaching of AI literacy:
- Learning by doing beats learning by telling. Abstract discussions of AI capabilities don’t compare to using these tools for meaningful, authentic work.
- Modeling matters. Even tech-savvy students benefit from seeing effective AI collaboration demonstrated before trying it themselves.
- Critical evaluation is crucial. Beyond being critical of the AI output itself, students need explicit encouragement to prompt AI for critical feedback about the work that’s generated so that they can avoid over-emphasizing supportive responses.
- Discussion complements experience. Making time for students to share discoveries, strategies, and ethical considerations dramatically enhances learning.
- Many students are only scratching the surface. Despite growing up with technology, many students are using AI at only the most basic level.
Looking Forward
I’m already planning improvements for the next iteration, if I get invited to teach it. I’d like to expand to six sessions to allow more time for pitch preparation, add more structured modeling of effective AI interactions, and incorporate explicit critical evaluation throughout the process.
Beyond these specific changes, this experience has reinforced my belief that the ability to use AI through collaborative conversation will be a significant differentiator for students entering college and the workforce. Those who understand how to use AI as a genuine thought partner — rather than just a search engine or content generator — will have a meaningful advantage.
This doesn’t mean uncritical acceptance of AI tools. Some students expressed valid ethical concerns about AI use, and these discussions were valuable parts of the learning process. What’s important is that they make informed decisions based on actual experience rather than surface-level understanding.
In the end, the most powerful AI literacy comes from a combination of authentic experience and reflective conversation. Young people need opportunities to use these tools for real, meaningful work, coupled with space to discuss what works, what doesn’t, and what it all means for their future.