NPCs as People, Too: The Extreme AI Personality Engine

📝 Abstract

PK Dick once asked “Do Androids Dream of Electric Sheep?” In video games, a similar question could be asked of non-player characters: Do NPCs have dreams? Can they live and change as humans do? Can NPCs have personalities, and can these develop through interactions with players, other NPCs, and the world around them? Despite advances in personality AI for games, most NPCs are still undeveloped and undeveloping, reacting with flat affect and predictable routines that make them far less than human–in fact, they become little more than bits of the scenery that give out parcels of information. This need not be the case. Extreme AI, a psychology-based personality engine, creates adaptive NPC personalities. Originally developed as part of the thesis “NPCs as People: Using Databases and Behaviour Trees to Give Non-Player Characters Personality,” Extreme AI is now a fully functioning personality engine using all thirty facets of the Five Factor model of personality and an AI system that is live throughout gameplay. This paper discusses the research leading to Extreme AI; develops the ideas found in that thesis; discusses the development of other personality engines; and provides examples of Extreme AI’s use in two game demos.

💡 Analysis

PK Dick once asked “Do Androids Dream of Electric Sheep?” In video games, a similar question could be asked of non-player characters: Do NPCs have dreams? Can they live and change as humans do? Can NPCs have personalities, and can these develop through interactions with players, other NPCs, and the world around them? Despite advances in personality AI for games, most NPCs are still undeveloped and undeveloping, reacting with flat affect and predictable routines that make them far less than human–in fact, they become little more than bits of the scenery that give out parcels of information. This need not be the case. Extreme AI, a psychology-based personality engine, creates adaptive NPC personalities. Originally developed as part of the thesis “NPCs as People: Using Databases and Behaviour Trees to Give Non-Player Characters Personality,” Extreme AI is now a fully functioning personality engine using all thirty facets of the Five Factor model of personality and an AI system that is live throughout gameplay. This paper discusses the research leading to Extreme AI; develops the ideas found in that thesis; discusses the development of other personality engines; and provides examples of Extreme AI’s use in two game demos.

📄 Content

KEYWORDS artificial intelligence, personality, non-player characters, video games

ABSTRACT PK Dick once asked “Do Androids Dream of Electric Sheep?” In video games, a similar question could be asked of non-player characters: Do NPCs have dreams? Can they live and change as humans do? Can NPCs have personali- ties, and can these develop through interactions with play- ers, other NPCs, and the world around them? Despite ad- vances in personality AI for games, most NPCs are still undeveloped and undeveloping, reacting with flat affect and predictable routines that make them far less than human— in fact, they become little more than bits of the scenery that give out parcels of information. This need not be the case. Extreme AI, a psychology-based personality engine, creates adaptive NPC personalities. Originally developed as part of the thesis “NPCs as People: Using Databases and Behav- iour Trees to Give Non-Player Characters Personality,” Extreme AI is now a fully functioning personality engine using all thirty facets of the Five Factor model of personali- ty and an AI system that is live throughout gameplay. This paper discusses the research leading to Extreme AI; devel- ops the ideas found in that thesis; discusses the develop- ment of other personality engines; and provides examples of Extreme AI’s use in two game demos.

INTRODUCTION: THEN AND NOW on-player characters include all the inhabitants of the game world who aren’t being played by human beings; they are, in effect, the original virtual actors. Their roles include everything from the bit parts (man in crowd, wom- an in shop) to main characters (player’s confidante, love interest, main villain), yet they rarely have any autonomy or development and do not react to player actions or world events as real people would. At best, they change as the script dictates. At worst, they merely repeat the same tidbit of information again and again, no matter that the world around them is going down in flames. The original 2011 project (unnamed at the time; referred to in this paper as the prototype personality engine, or PPE) sought to change this, allowing NPCs to be “utilised much more effectively and realistically—made, in fact, more hu- man by giving them personalities that change over time and through interactions with the players, other NPCs, and the game world,” giving them “added realism [that] would add depth and flavour where before there was only cardboard two-dimensionality” (Georgeson 2011). The objectives of the project included (as stated in the original):

 creating support NPCs whose personalities devel- op realistically over time depending on interac- tions with a player in a game environment
 comparing these NPCs to a set of unchanging NPCs whose reactions are controlled by random chance

As the PPE was developed into Extreme AI (ExAI), the system was further refined and developed. Its primary ob- jectives became:

 creating more potential interactions for developers to call (originally there were but three—kindness, annoyance, and intimidation—which worked well for the PPE but were far too limiting for realism)  speeding up the system so that complex interac- tions didn’t noticeably slow down action in a game  moving away from using a database for storage so that web- and mobile-based games could utilize the system

This paper summarizes the research and original work on the PPE, personality engine development since that time, and the current capabilities of ExAI. PERSONALITY ENGINES AND ACADEME While limited in scope, the PPE was, at the time, the only one of its kind. Given increases in computational power, AI could be (and sometimes is) provided with increased re- sources, but it is still given short shrift compared to graphics (Doherty & O’Riordan 2006; Lemaitre, Lourdeaux, & Chopinaud Jan 2015). As stated in 2011:

For example, Final Fantasy XII (Square Enix 2006) NPCs look very realistic, but they react in ways similar to such characters in the first Final Fantasy games in the 1990s, with possibly only a very basic finite state ma- chine guiding the characters, and this is the same in role- playing games generally: When the player approaches the NPC for the first time, the character says something about the red dragons guarding the Great Treasure—and repeats the same thing every time the player comes near. After the player has completed this quest and returns to tell the NPC about it, the NPC may (or may not) be com- plex enough to say something about being happy no longer having to deal with the dragons. Jeffrey Georgeson Christopher Child Quantum Tiger Games, LLC City University London 3431 W 17th Ave Northampton Square Denver, CO 80204 USA London EC1V 0HB, UK E-mail: jg@quantumtigergames.com E-mail: c.child@city.ac.uk

NPCS AS PEOPLE, TOO: THE EXTREME AI PERSONALITY ENGINE N

Usually even this sort of thing does not iterat

This content is AI-processed based on ArXiv data.