Foundations
My main goal for Milestone 1 was designed to be all about building a foundation of understanding for myself on the current state of generative AI in gamespaces; brick by brick.
For this research based milestone, I came up with a list of core and sub-questions I determined necessary for me to answer before I could progress though my plan to create an intelligent NPC prototype.
Milestone 1 Research Questions
- Understand the current state of generative AI in games
- 1a. Generative AI Models being used in game environments
- 1b. Studios/games utilizing generative AI
- Identify the difference between generative AI and traditional Game AI
- 2a. Specific language for each
- 2b. Key differences
- Document the current mood on generative AI use in games
- 3a. Available opinions and thoughts on arguments; pros/cons
- 3b. Draft my own hypothesis on generative AI use in games
- Research current tools of implementing generative AI into game engines.
- 4a. Specific generative AI tools for games-what exists?
- Identify what are “Intelligent” NPCs?
- 5a. Is this a defined term? What else would they be called?
- 5b. How are they made in game engines?
- 5c. What games have done it?
- 5d. Are there any step-by-step guides?
Research
Documentation of Findings
As I conducted my research, I realized the subject I was diving into was massive in terms of the current tools available, the concepts developers have tried and not tried, as well as industry and non-industry opinions.
I took extensive notes trying to wrap my head around it all, and ended up with a messy 40 page document of notes!
I have found through experience that if you do not worry about formatting on a first draft of anything from creative writing to research papers, allowing that flow state does not inhibit your motivation, creativity, or work you can produce. I had a lot of good information there, I just needed to organize it neatly.
I was then able to take that research and reformat my notes into a neater document:
My Findings
I summarized my findings for my first Milestone presentation. I gave my presentation to my peers in my MFA cohort on Wednesday September 17th, 2025.
My general understanding of the current state of generative AI in the game industry has evolved.
The “AI Revolution” is upon us. According to several studies, most (<90%) game developers are already using generative AI in some form of their workflow and also believe it will become core to future game development
Generative AI in gamespaces is a bit of a wild west at the moment. It is such a young concept (less than 5 years old) that not many studios have yet implemented it into game play. However, since it is so new, there is a massive opportunity of potential for new ideas of how to utilize generative AI in gamespaces to those who wish to explore it. Now is the time!
Generative AI means many things (it has many sub-categories and different language model types). I am mainly interested in LLMs (large language models) and SLMs (small language models) for my project.
A LLM will be the key to my intelligent NPC prototype. However, from what I discovered about SLMs, they may be the key to alternative ideas I have on how to implement generative AI into game play through other entities and methods. More to come on this…
Gen. AI is an exciting new turn of technology. This project has the potential to evolve because generative AI as a whole has the potential to evolve. Now that I feel I have met my Milestone 1 goals, ideas are churning in my brain on never-before-thought of possibilities that we as game developers could create, with humans at the helm conducting this technology within the gamespace.
Next Steps
Milestone 2
The main goals of my Milestone 2 are Workspace Set Up & NPC Pre-Development.
This entails these priorities:
- Setting up a space in a game engine (depending on which plugin I choose to use)
- Implement the genertaive AI plgin within the game engine
- Set up the game scene, NPC sprites/assets
- Define the “scope” of the prototype
- Goal of player in this prototype -> what is the “situation”? What are they communicating with the NPC about?
- The personality and bounds of the NPC
I think this is a solid plan, however…there are still some considerations I must make:
Do I need to make any alterations to the original milestone 2 goals?
Should I change the type of intelligent entity I am making in game engine with gen. AI?
Do I work with LLM or a SLM? Both? Which plug-in? Which engine?
What is the "situation" I will design for this prototype as well as the goals of the NPC and the players?
Do I instead switch gears to work on a design for a new type of "bounded language model" for intelligent entities within gamespaces?
I will spend the next few days mulling over the new ideas I have thought of during my research before I begin official work. There is much to think about!
Until next time,
~Lauren
