AI isn’t a completed product you either master or ignore. It’s a set of capabilities you can learn to use, and the fastest way to get good is to start small, be curious, and iterate. The following guidance is practical and classroom-centered: reduce administrative friction, apply systems thinking to design, and treat students as creators with AI.
Why start small
Begin with modest, high-value problems that currently waste teacher time or get in the way of instruction. Small wins create momentum and produce reusable components you can combine later. Examples that work:
- Automate a scheduling annoyance (e.g., translate a rotating “waterfall” schedule into a simple HTML app so staff stop decoding the calendar).
- Build a random name picker from a spreadsheet and a small script instead of relying on popsicle sticks or third-party apps.
- Create study tools or simple apps students can use and improve across iterations.
Focus on data and architecture first
AI delivers polished outputs quickly, but polished doesn’t mean useful. The single biggest practical barrier is messy inputs. If your spreadsheet headings, IDs, and naming conventions are designed for human readability only, models and scripts will struggle.
- Clean data: develop consistent IDs, clear column headings, and functional layouts that AI and code can consume.
- Make small, testable data sets before scaling up. If the output is wrong, the problem is usually upstream—either the data or the question being asked.
Use systems thinking to avoid “pretty garbage”
Systems thinking helps you break complex problems into manageable pieces and prevents premature conclusions. Don’t start with the graph you want; start with the question you actually need to answer, then map the inputs, outputs, and how they connect.
- Sketch the end-to-end flow. Breaking work into smaller modules avoids exceeding context limits in AI tools and reduces brittle outputs.
- Iterate: build one component, confirm it works, then connect components into larger systems that can reveal unexpected insights.
Make students creators with AI
Move beyond treating AI as a shortcut for assignments. Design projects where students teach, design, and build. Intergenerational and cross-role collaboration can be especially effective.
- Have students teach a concept to adults and then co-design a storyboard for an app that teaches that concept.
- Let students use AI-assisted coding tools to build simple study apps, then evaluate against success criteria and iterate.
- Use AI to surface how models work and why outputs may be flawed—learning comes from the design and revision cycles, not just the final product.
Leadership, culture, and professional learning
Leaders must be willing to be vulnerable and visibly experiment. Culture matters: people will follow if they see leaders trying, failing, and improving.
- Run short, focused PD or teacher-led sessions where colleagues can bring real questions and small projects.
- Encourage staff to reach out to peers and communities (online or international) to adapt ideas rather than trying to invent everything in isolation.
- Acknowledge the fatigue and anxiety staff feel; start with low-risk, personally meaningful AI tasks (trip planners, recipes, hobby projects) to build confidence.
Policy and pedagogy: opportunities, not just restrictions
Blanket bans or punitive policies only push use into the shadows. Instead, combine policy with education and modeling:
- Teach students responsible uses and how to collaborate with AI rather than simply policing it.
- Accept that AI will be part of students’ futures and prepare them to use it as a partner, not a shortcut to avoid learning.
- Use real classroom projects to examine academic integrity, prompt design, and revision history rather than relying solely on blocking technologies.
Practical next steps you can do today
- Pick one administrative pain point and build a tiny automation or spreadsheet that helps—then keep that component in a shared folder for reuse.
- Run a single intergenerational or cross-department project where students design an app or study tool and iterate twice.
- Audit one dataset in your school for cleanliness and consistent IDs; fix it and observe how much easier it is to build tools from it.
- Hold a low-stakes “fun AI” session: have staff use AI to plan a trip, swap recipes based on pictured ingredients, or solve a hobby problem to build comfort before moving to classroom use.
Start small, think big: small, well-designed experiments build the foundations—clean data, modular systems, and an inquiry mindset—that let AI increase meaningful learning time rather than simply creating prettier garbage. Lead by modeling curiosity, vulnerability, and iteration, and prioritize student creation over consumption.
AI Answers
How should a teacher who is new to AI begin?
Start with a small, practical problem that wastes your time—automate it or prototype a simple app. Use available tools (Claude, ChatGPT, agent tools) to create one reusable component and iterate from there.
What prevents AI outputs from being useful?
Messy inputs and unclear goals. Clean data structures, consistent IDs/naming, and a clear question or success criteria are essential to avoid polished but nonfunctional results.
How can leaders encourage teachers who are resistant or anxious about AI?
Model vulnerability by experimenting publicly, run short teacher-led PD, promote low-risk personal projects to build confidence, and create spaces to share small wins and failures.
How can students be engaged with AI in authentic ways?
Treat them as creators: have them teach concepts to non-experts, storyboard and build simple apps with AI assistance, evaluate against criteria, and iterate to learn both subject matter and how AI works.
For more context, listen to the original episode of Smarter Campus Podcast: Start Small, Think Big: Matthew Ignash on Experimenting with AI in Education.