The Idea Collision Method: How to Use AI and a Concept Combination Matrix to Unlock Your Next Big Idea

Show notes

A practical how-to guide for business writers, educators, and anyone who’s ever stared at a blank page.

You know that feeling when you need a fresh idea, a new angle for a blog post, a creative lesson plan, a different way to approach a problem, and your brain just keeps serving up the same tired options? Yeah. That’s not a creativity problem. That’s a collision problem.

The best ideas almost never come from staring harder at the thing in front of you. They come from smashing two unrelated things together and seeing what survives the impact.

That’s the core principle behind the Concept Combination Matrix, a simple but surprisingly powerful technique for generating novel ideas by forcing connections between things that would never normally meet. And when you pair it with AI, the whole process gets faster, stranger, and a lot more productive.

Here’s exactly how to do it.

What Is a Concept Combination Matrix?

At its simplest, a Concept Combination Matrix is just a grid. One set of ideas runs along the rows. Another, completely different set runs along the columns. Every place a row and column intersect becomes a prompt: what would happen if these two things collided?

The technique has roots in lateral thinking and structured creativity methods. The key insight, and the reason it actually works, is that the combinations are forced. You can’t opt out of the weird intersections. The row that seems irrelevant, the column keyword that feels absurd, those are exactly the ones that tend to crack something open.

Think of it like this: a 5-row by 5-column grid gives you 25 intersections. Even if 20 of them feel like dead ends, the other 5 might contain ideas you’d never have reached by thinking directly about the problem.

Quick Visual Rows = your domain (e.g., teaching methods, content formats, product features). Columns = random keywords (e.g., volcanoes, jazz, origami). Every intersection = one new idea to explore.

Where AI Comes In

Traditionally, you’d fill in every cell of the matrix yourself, which is both exhausting and prone to the same mental ruts you were trying to escape. Your brain wants to take shortcuts. It wants to make the intersections make sense in familiar ways.

AI doesn’t have that problem. It doesn’t have your habits, your assumptions, or your professional blind spots. Ask it to combine ‘formative assessment’ with ‘shipwrecks’ and it won’t hesitate or self-censor. It’ll generate something specific, named, and fully articulated, often something you’d genuinely never have thought of.

Here’s what AI specifically adds to the matrix method:

1. Speed. AI generates 30 ideas in the time it would take you to fill in 5 cells manually.

1. Strangeness. Without your mental habits filtering ideas before they’re born, the AI goes places you’d unconsciously rule out.

1. Specificity. You don’t get vague gestures toward a concept, you get a named idea with a description and a use case.

To run this with an AI tool, the prompt is simple. Give it your rows (the domain you’re working in), your columns (the random keywords), and ask it to generate a named idea plus a one-sentence description for each intersection. That’s it. The whole matrix can fill up in under a minute.

Prompting Tip Use genuinely random keywords, not ones that feel adjacent to your problem. The stranger the keyword, the more interesting the collision. Try a random word generator, open a dictionary to a random page, or just ask the AI to surprise you.

Example 1: Business Writing

Let’s say you’re a content strategist who writes for a B2B software company. You’ve been producing the same types of content for months, case studies, how-to guides, the occasional thought leadership piece, and everything is starting to feel a bit formulaic. You need fresh angles.

Here’s how the matrix would look:

1. Rows: your content formats, case study, how-to guide, opinion piece, data report, interview

1. Columns: Ask AI for five random keywords, beekeeping, shipwrecks, origami, jazz, volcanoes

You run that grid through AI and 25 ideas come back. Most are interesting. A few are genuinely exciting. Here are two worth developing:

How-to Guide × Origami → “The No-Going-Back Guide to Product Decisions”

The insight from origami is that every fold is permanent, once you crease the paper, that decision is locked in. The AI uses that metaphor to suggest a business how-to post about making irreversible strategic decisions carefully: how to know when a choice is a fold, how to plan before you commit, and how to work beautifully within the constraints created by past decisions.

This is a genuinely useful piece of business writing. And it has a memorable, distinctive frame that no competitor is likely to have already published, because nobody would have reached for origami as a metaphor for product strategy without the forced collision of the matrix.

Data Report × Jazz → “Improvising with Imperfect Data”

The jazz connection opens up a completely different argument: that great data analysts, like jazz musicians, don’t wait for perfect conditions before they perform. They improvise brilliantly within constraints. They work with incomplete information. They make decisions in real time and adjust.

Again, sharp, specific, differentiated. This isn’t a piece that would come from a content brief that said ‘write something about data analysis.’ It came from a forced collision with an unrelated concept.

Key Takeaway The random keywords didn’t distract from business relevance, they created it, by forcing a fresh metaphorical frame onto familiar material.

Example 2: Teaching Methods with AI

Now let’s try the same technique in an education context. Imagine you’re a teacher or instructional designer exploring how to integrate AI more creatively into your classroom practice. You’ve read the usual articles. You know AI can help with lesson planning and differentiation. But you want ideas that are genuinely new, not just repackaged versions of what everyone else is doing.

The matrix this time:

1. Rows: AI-assisted teaching approaches, personalized learning, formative assessment, lesson co-creation, Socratic questioning, project-based learning

1. Columns: the same five random keywords, beekeeping, shipwrecks, origami, jazz, volcanoes

Two intersections that immediately stand out:

Formative Assessment × Shipwrecks → “Black Box Recovery”

Flight recorders, black boxes, capture everything that happened before a crash. The AI uses that frame to suggest a formative assessment approach where, after a failed quiz or assignment, students don’t just get a grade. They work with AI to reconstruct the exact decision points in their thinking that led to the wrong answer.

Instead of ‘you got question 5 wrong,’ students and teachers explore: where did the reasoning go off course? Which conceptual fold was the bad one? The AI logs the thinking process throughout the original task and helps students trace back through it. Mistakes become data rather than verdicts.

This is pedagogically sound, it’s directly aligned with metacognitive learning research, and it’s also a vivid, memorable frame that a teacher could actually explain to students in a way that sticks.

Lesson Co-Creation × Jazz → “The Fake Book for Teachers”

Jazz musicians use what’s called a ‘fake book’, a lead sheet that gives you the melody, the chord changes, and the key, but leaves all the interpretation to the performer. The AI uses this to suggest a lesson design approach where the AI generates the core structure of a lesson (objectives, key turning points, entry and exit tasks) but deliberately leaves white space for the teacher to improvise in the room.

This is a subtle but important idea. Most AI-generated lesson plans are over-specified, they tell teachers what to say, when to pause, how to respond. That kills the instinct that makes great teaching great. The fake book model does the opposite: it gives structure without scripting, so the teacher’s responsiveness to the actual room is preserved.

Key Takeaway The matrix didn’t produce gimmicks, it generated pedagogically grounded ideas with real classroom application, by approaching familiar teaching concepts from a completely unexpected angle.

How to Run Your Own Session

Ready to try it? Here’s the exact process, start to finish:

1. Define your domain.

What are you actually trying to generate ideas about? Be specific. ‘Marketing’ is too broad. ‘Fresh angles for our Q3 email newsletter’ is just right. Your rows should reflect the specific facets of that domain, content formats, teaching approaches, product features, whatever fits.

1. Generate random column keywords.

Use a random word generator, open a physical dictionary, or just ask your AI tool to give you five surprising words. Resist the urge to pick keywords that feel relevant. Relevant keywords produce obvious intersections. Strange keywords produce interesting ones.

1. Give the grid to AI.

Paste your rows and columns into your AI tool of choice with a clear prompt: ‘Generate a named idea and a one-sentence description for each intersection in this matrix.’ Done. The whole grid fills up in under a minute.

1. Read without evaluating.

Scan all 25 (or however many) intersections before you judge any of them. Your first reaction to most cells will be ‘that’s weird’, that’s fine. Keep reading. The goal at this stage is exposure, not selection.

1. Star the ones that feel alive.

You’ll notice 3 to 5 intersections that create a small spark, you might not even be able to explain why yet. Star those. Don’t overthink the selection.

1. Develop the starred ideas.

Go back to AI and ask it to develop each starred idea into a full paragraph, a concept brief, or a draft outline, depending on what you’re building toward. This is where the ideas start to become real.

One More Tip The whole session, from blank matrix to 5 developed ideas, takes about 20 minutes with AI. Plan a short debrief at the end to decide which ideas are worth pursuing and which need another round of collisions.

The Takeaway

Novelty doesn’t come from thinking harder about the problem directly in front of you. It comes from collision, from forcing your domain to interact with something it has no business interacting with, and seeing what sparks.

The Concept Combination Matrix gives that process structure. AI gives it speed and strangeness. Together, they turn a frustrating blank-page moment into a 20-minute session that ends with ideas you genuinely wouldn’t have reached any other way.

One thing worth remembering: AI doesn’t replace your creative judgment here. It dramatically expands the surface area of what you get to judge. You still decide which ideas are worth developing. You still bring the context, the taste, and the understanding of what your audience actually needs. The AI just makes sure you’re not limited to the ideas you could have predicted.

So the next time you’re stuck: open a dictionary, pick five random words, build a grid, and let the collisions do their work.

Try it today with any problem you’re stuck on, and share your most surprising intersection in the comments.

Note: I used this myself to look for ideas for future podcasts. The rows are the phases or aspects of the creative process, and columns are random words.