AntyVersum

    Making games with passion
Human Addiction Behaviours As The Case Study for AI #25
Slowly approaching the stage of development, where AI will be taken into workshop.
Want to help? Whishlist us on Steam (With Pre-Alpha Experimental Playtest Demo)

Slowly approaching the stage of development, where AI will be taken into workshop.


There is many ways to approach AI. But early The Sims series executed it very well. So it is worth base on that, with support of modern technologies. In principles, base mechanics will requires needs and corresponding action of characters, to sustain such needs. For example hunger, need for sleep, bladder, or socializing. Here Maslow's piramide hierarchy of needs is a good reference.


But I want to achieve a bit more than just that. In the end I got whole community living. Such community can potentially create own sub groups and norms. Or whole community can change their own norms. Like what is allowed and whats not. What is perceived as good, or bad. Kenshi, or Rimworld games are good reference points here.


Such mechanics leads to interesting case studies.

If going into very basics, like using chair, technically there are no limitations, or obstructions, to which character can use as such. Unless is the technical limitation. Like the small child, which is unable to climb up.


We can enter here with a chair into cultural and historical behaviours, where like in Japan for example, sitting without chairs is not something unusual. From the point of view of AI, such behaviour is driven by ethics, trends, needs, community perceptions, prohibitive regulations etc. Furthermore, these can vary a lot between friends, family, in the public, or at work. For example work palce can enforce code of conduct, or dress code. While among close friends, such rules can be completely loose.


Analyzing for years human behaviorus and personal experience, I was investigating, how such behaviours can be reflected into a game AI, for dynamic evolving environment. What makes one, or other person do things, that others don't. This is big subject on its own, but here are few interesting observations, rules and points, that can affect actions, taken by the person, or character in case of AI.


Side note, talking about AI, I am not talking about ChatGTP like AIs and generative tools. But AIs that are used in general in games.


For that, addictive behaviours and items was taken as a case study, as it is very interesting subject.

What makes us to make do a thing. Either needs (fulfill needs), requirements (someone tells you to do thing), or enjoyment (go chat, or play). People do prioritise and tend to do things that they enjoy. It is quite simple in principles. But lets say analyse cigarettes. for some it is enjoyable, for others is distracting and pushing away. If we consider base entry point as an enjoyment, everyone would like to try smoke. Among friends, that even more encouraging. It was extremely popular behaviour in halve of last century. Was cool. And it is still considered cool among many groups. So we increase enjoyment level, by adding coolness.


What if we now start applying prohibitive factor. We reduce that drive to try smoking. Not necessarily stopping completely. But person with relevant personalities, may be affected by such factor more than other. Then we apply knowledge of lets say health. Educating about side effect, can reduce further factor of willing to try to smoke. But exploring trait, still may want to try it. Experience of someone smoking and smelling bad, can further push away, from willing to try.


We can add various factors to such actions and make such more or less willing to be executed by AI. Just like in the real life.


Now if we extend addition case study into various behaviours, we suddenly creating complex and dynamic environment, while applying relatively simple rules. We can affect characters how to eat, clean, work, or even how to sleep. As well, as how to behave around others. What to talk and not talk about. Like gained an experience of each characters.


There is many levels of technicalities here. But idea is, to build up basics rules, around which whole system revolves. Such may be enough, to build very dynamic and often unexpecting evolving ecosystem. See here for example Factorio, or Minecraft. These games have basics rules for game mechanics. Adding small variations, creates whole dynamic complexity.


Such idea is, that AI community can affect individuals. As well, as individual by taking actions, can affect whole community. However, end results and actual executions are yet to be seen in the game. 😎

Reddit Discussion

Leave a Reply

Your email address will not be published. Required fields are marked *

Scroll to Top