Brainssssss: Zombie Behaviour Tech | Sarah Dobie | NZGDC 2025
Keywords
Summary
182 words
Critical Evaluation
Value of the Information & Strength of the Argument
The talk provides valuable insights into practical game AI implementation, covering a comprehensive range of systems from low-level sensing to high-level decision-making. The argumentation is solid, as each system is explained with its purpose and how it contributes to the overall AI behavior. The use of real-world examples from the game, such as noise propagation and queueing, demonstrates the effectiveness of the approach. The speaker’s experience and the coherent architecture lend credibility to the content. The live demo further reinforces the practicality of the tools, showing how they can be used to create complex behaviors efficiently.
Scientific Rigor, Source Quality, Title Accuracy
The talk is based on the speaker’s direct experience developing AI for a commercial game, which provides a strong foundation of practical knowledge. However, no external sources are cited, and the content is presented as a personal overview rather than a peer-reviewed study. The title accurately reflects the content, focusing on zombie behavior tech. The talk is well-structured and technically rigorous, with clear explanations of the systems and their interactions. The lack of external references is a minor limitation, but the depth of detail and internal consistency compensate for it.
201 words
Title / Content Match
The title 'Brainssssss: Zombie Behaviour Tech' accurately reflects the content, which focuses on the AI systems behind zombie behavior in a game.
Quality & Reliability
8/10
The talk is a detailed technical overview by a senior programmer with 7 years of experience, presenting a coherent and well-structured AI architecture. The content is based on practical implementation in a commercial game, with clear explanations of systems and their purposes. No external sources are cited, but the technical depth and internal consistency support a high reliability score.
Key Moments
Markers derived by PSI from the transcript: the creator did not define chapters.
- Introduction to the talk and the game 'Into the Dead: Our Darkest Days'.
- Overview of the AI brain architecture: low-level, mid-level, and high-level brains.
- Explanation of the low-level brain: sensors for sight, noise, touch, and damage.
- Mid-level brain: stimulus entity tracking, triage scoring, and alertness states.
- Targeting system and temporary targets for obstacles.
- High-level brain: blackboard, AI tasks, hierarchical finite state machines, and behavior trees.
- Physical action: AI uses the same character systems as the player.
- Additional systems: pathfinding web, noise propagation, queueing, target sharing, and territory.
- Live demo of AI tools in Unity, showing state machine and behavior tree creation.
Contribution & Novelties
The talk provides a comprehensive and practical overview of AI systems for a 2.5D game, with a focus on modularity and reusability. The use of a hierarchical finite state machine combined with behavior trees, and the integration of a custom pathfinding web, are notable contributions. The talk also highlights the importance of sensory realism and fairness in stealth games.
Pour aller plus loin :
- Hierarchical finite state machine — A foundational concept for structuring complex AI behaviors.
- Behavior tree — A widely used AI architecture for game development.
- A* search algorithm — The algorithm used for pathfinding in the custom web system.
102 words
Radar Profile
The radar profile shows high scores in quantity and quality of information, with a slightly lower technical level, indicating a well-balanced talk that is informative and accessible. The reliability score is also high, reflecting the speaker's expertise and the practical nature of the content.