itch.io

I'm here to share my project, yes, but mostly just hoping to find people who might enjoy it. Not trying to spam!

Hey everyone! 👋

I'm about three months into a pretty wild journey: learning AI game development from absolute zero. No CS degree. No coding background. Just curiosity and a stubborn belief that I could figure this out.

Why I'm posting here

I've been lurking in the itch.io community for a while, reading dev logs, downloading jams games, and quietly absorbing. You all seem to be the kind of builders who actually make things; and I could really use some community wisdom as I hit the messy middle of my first real project.

What I'm building

I'm working on a small prototype where NPCs learn from player behavior instead of just following scripted paths. Think: enemies that adapt to your combat style, characters that remember how you treated them. Nothing revolutionary by AAA standards, but for someone who started three months ago thinking "behavior tree" was something in my backyard? It feels like magic.

Three things that surprised me so far:

1. AI in games isn't just about "smart" enemies - It's about making worlds feel alive. An NPC that flinches when you draw a weapon (because it "learned" you're aggressive) creates more immersion than any fancy texture ever could.
2. You don't need to be a math genius - I kept avoiding this because I assumed I needed calculus and linear algebra. Turns out, modern tools (Unity ML-Agents, Unreal behavior trees) let you see what's happening. The math becomes something you learn when you need it, not before you can start.
3. The "aha" moment is real - Watching an agent you trained finally figure out how to navigate toward a goal? Ridiculously satisfying. Like teaching a digital pet a trick.

Where I'm stuck right now

I'm hitting the classic beginner wall: my training environments feel janky, and I'm not sure if my agents are learning "badly" or if I've just set up the rewards wrong. Would love to hear from anyone who's trained reinforcement learning agents—how do you debug when the AI learns the wrong behavior?

Also curious: for those of you building games with AI systems, what's one thing you wish you'd known when you started?

Why I'm extra motivated

Full transparency: I'm documenting this journey because I'm backed into a corner—I found a course that's teaching me all this, but it's a Kickstarter project that needs support to keep going. So I'm learning in public, sharing what works (and what fails), and hoping the community finds some value in the process.

If you're curious, the course is called AI Powered Game Dev For Beginners - it's the reason I'm not completely lost right now.

But honestly? I'm here first because I want to learn from people who've actually shipped games. If you've got advice, critiques, or just want to point me toward resources that helped you, I'm all ears.

Thanks for reading, and maybe I'll see some of you in the jam scene once this prototype stops being embarrassing. 😅