AI Lesson Planning for Teachers: What Actually Works (and What Doesn't)

Every teacher has tried asking ChatGPT for a lesson plan. Most of them got something that looked impressive for about 30 seconds and then fell apart the moment they tried to actually use it.

The output was generic. It didn't know your students. It taught to no one in particular. And the "discussion questions" were the kind of questions no real student has ever asked.

AI lesson planning can be genuinely useful — but only if you understand what it's actually good at and how to work with it rather than around it.

What AI is genuinely good at

Generating structure fast. A solid lesson plan has a repeatable architecture: opening hook, core content, discussion, application, close. AI can scaffold that structure in seconds, leaving you to fill in the substance. That's legitimately useful.

Breaking a topic into teachable units. If you know you want to teach on "prayer" for six weeks, AI can quickly map out six distinct angles — what prayer is, why it's hard, different types, what happens when God seems silent, praying together, etc. It sees the landscape faster than most teachers can.

Generating discussion questions in volume. Not all of them will be good. But if you need 20 questions to choose 6 from, AI is faster than your brain at 11pm on a Saturday.

Drafting take-home worksheets. The worksheet format is predictable enough that AI handles it well: key point, reflection questions, action step.

What AI consistently gets wrong

It doesn't know your students. This is the core failure. A lesson plan for "teenagers" is a lesson plan for no one. Your teenagers — their age, their questions, their level of faith, their specific doubts, what they thought was funny last week — none of that is in ChatGPT's context. Generic input produces generic output.

It tends toward the safe and obvious. Ask for a lesson on forgiveness, and you'll get the prodigal son. Ask for one on doubt, and you'll get Thomas. The AI defaults to the most recognizable example, not necessarily the most useful one for your group.

It doesn't understand how long things take. A "15-minute opening activity" it suggests might take 4 minutes or 45 depending on your group. Timing is one of the hardest things to teach to an AI, and generic lesson plans routinely underestimate transitions, discussion sprawl, and setup time.

It loses the thread across a series. If you ask for six lesson plans in six separate conversations, you get six lessons that don't know about each other. There's no arc, no building, no callbacks to earlier sessions.

How to actually get useful output

The secret is context upfront. Don't ask for a lesson plan — give the AI everything it needs to write for your room:

• Who are your students? (Age, background, where they are in their faith)
• How long is your session?
• What do you want them to walk away understanding, feeling, or doing?
• What's your teaching style — discussion-heavy, story-driven, direct instruction?
• Any constraints — things to avoid, vocabulary to use or not use, resources you'll have available?

That context transforms the output. Instead of a lesson for no one, you start getting something that at least knows what room it's being taught in.

The second thing that helps: treat AI output as a first draft, not a finished product. The structural scaffold is usually good. The specific language, examples, and questions almost always need editing to sound like you.

Where AI tools differ

Most AI lesson planning involves pasting context into a general chat tool and hoping for a useful response. That works, but you're doing a lot of the work — knowing what to ask, how to structure your input, and how to evaluate what comes back.

Dedicated tools like PopLesson are built specifically for this. Instead of a blank chat window, ForgeAI asks you the right questions, builds a session-by-session outline for your approval first, then generates the full curriculum — teaching guide, discussion questions, opening activity, student worksheet — in one pass.

The difference is that the context isn't just a prompt — it's the whole point. The AI knows it's building a 6-session Sunday school series for 9th and 10th graders at a small evangelical church, not "a lesson plan."

The honest verdict

AI lesson planning saves real time if you use it right. The hours you'd spend building structure, brainstorming questions, and formatting worksheets — AI handles that. The judgment calls about what to emphasize, what example will land with your specific kids, and what your group needs to hear right now — that's still yours.

Think of it as a very fast first draft and a thinking partner, not a replacement for knowing your students.

The teachers who get the most out of AI tools are the ones who bring the most to them: clear outcomes, real context about their group, and the editorial judgment to know which outputs to keep and which to throw away.