#gameai — Public Fediverse posts

Interesting video: "Hierarchical Dynamic Pathfinding for Large Voxel Worlds - GDC Vault" https://www.youtube.com/watch?v=pG9SoTF1mCI

#gamedev #gameai #programming

Interesting video: "Pathfinding in games. How to do it for 30k entities." https://www.youtube.com/watch?v=anGdYJu_eH4

#gamedev #gameai #programming

Game AI Uncovered: Volume Five is out 🎉

I must say, the cover art is extremely slick -- it's from Lords of the Fallen. Hopefully the content is as well 😊 It features one chapter from your humble servitor, titled AI LOD in Multiplayer Games, alongside many others from great contributors.

It's available here to order: https://www.routledge.com/Game-AI-Uncovered-Volume-Five/Roberts/p/book/9781032840536

#GameDev #GameAI

Interesting read: "New pathfinding algorithm - Factorio Friday Facts" https://factorio.com/blog/post/fff-317 & "Belts optimization for 0.15" https://www.factorio.com/blog/post/fff-176

#gamedev #gameai #programming

Ladies and gentlemen, here's the story of the craziest bug I worked on during the late stages of shipping #MindsEye. The final fix went in exactly one week before release, just in time for the day 1 patch.

I wish I had the opportunity to do a full post mortem of that project, but this will probably never happen. That bug was a gem though. The world needs to know about it before my memory fades away.

#GameDev #GameAI #Programming

1/n

AMD Schola v2 drops today! 🔥

Our #opensource #UnrealEngine 5 plugin makes it easy to create AI that learns by playing your game — just like a human would.

Perfect ✅ for:

🚀 Smarter NPCs
🎯 Realistic opponents
🌍 Dynamic game worlds

https://gpuopen.com/amd-schola/?utm_source=mastodon&utm_medium=social&utm_campaign=schola

#GameDev #GameAI

My favourite way to play “Shooters” has always been co-op.
I switched my AI coding (Climate modelling, Chess, etc.) to making “bots” for the Quake Engine based games back in the 1990s, just so me and a few friends could have some fun against Artificial Enemies. It really took off.

For an arcade machine project in 1999, I had to invent “Artificial Stupidity”, because my Bots were too good for the players to beat. Playtesters died all the time, so I taught the bots to do things like:
⭐️Go the wrong way.
⭐️Miss valuable pickups.
⭐️Worsen their aim.
⭐️‘Hear’ and ‘see’ less.

Fun experience, teaching them to play like a human. As players progressed through the levels, the bots grew more intelligent, gradually providing more challenge, so players felt they got their money’s worth.
#coding #ai #gameai #shooters #coding #quake #coop #jollycooperation

Predicting Combat Outcomes in Total War
Miguel Lopez-Bachiller & James Kwan (Creative Assembly) reveal how they used supervised learning to improve AI battle decisions and autoresolve in Total War. Get insights from data, dev challenges & R&D collab.
📍 Nov 3–4 | Goldsmiths, London
🎟️ https://www.tickettailor.com/events/gameaieventscic/1651469
#GameAI #MachineLearning #TotalWar

🎮 When Research Meets Release Dates: Production-Grade RL for Games
Wesley Kerr (Riot Games) shares real-world lessons on making reinforcement learning work at production scale—beyond the hype.
🗓️ Nov 3–4 | 📍 Goldsmiths, London
🎟️ Get tickets: https://www.tickettailor.com/events/gameaieventscic/1651469
#GameAI #ReinforcementLearning #AIDevelopment

From Metrics to Meaning: Effort-Aware and Explainable AI for Game Pathfinding by Yetunde Longe-Folajimi (Wentworth Institute of Technology), shows how AI can help decode level difficulty & design smarter procedural content.

🎟️ Nov 3–4: https://www.tickettailor.com/events/gameaieventscic/1651469

#GameAI #ProceduralDesign #GameDev

Big Brain, btw, is designed to be data-oriented, so I might provide an API that lets you define game AI in .kdl files that look like this. It could also, for example, eventually use a hypothetical node editor (thing Blender Geometry Nodes), to define the same data.

#KDL #Bevy #BigBrain #GameAI

Seeking feedback from Bevy and Big Brain users: how does this new API look?

Under the hood, this would be a high-performance, low-overhead API powered by Observers, but this is all you would actually need to write yourself.

I'm pretty sure I can get this to work with Bevy's current featureset, including the Very Fun async actions situation here.

#Bevy #BigBrain #GameAI #Rust #RustLang