Tripleopolis: How I used AI to create a game and discover that I’m not entirely obsolete (yet)
Apr 13, 2023
Unless you’ve been living under a rock for the last few months you’ll no doubt have heard of the rapid advancements in the field of AI (artificial intelligence).
As a self-employed designer and developer this technology has the potential to radically impact my career in both disciplines. Since ChatGPT first launched last November I’ve been keeping a close eye on the AI revolution so see whether it;s more likely to open up new opportunities for me, or frankly to make my skill set somewhat redundant.
As I’m someone who learns far more effectively from practice rather than theory, I did what I always do when trying to learn something new – came up with a side project. a PWA-based game called Triplopolis.
If you’ve not heard of a PWA it stands for ‘Progressive Web App’, essentially a website that looks and functions like an app.
For this project I focused on the two major players in the AI space; ChatGPT and MidJourney. The AI platform ChatGPT has received most of the headlines and from a development standpoint ChatGPT can write effective, clean, production-level code in seconds. On the other hand AI image-generators such as MidJourney can create just about anything you can imagine, from photorealistic imagery to illustrated game assets and everything in-between.
Triplopolis at its core is a ‘match-3’ game where players combine elements to build their city – merging three or more trees creates a house, three houses makes a mansion, and so on… The concept has been in the back of my mind for years, as I’ve wanted to combine the aspects of much-loved games TripleTown and 1010 into a new concept whilst sprinkling in a few other twists along the way.
The project idea allowed me to test AI with a wide range of disciplines that I have at least a little knowledge of. Without AI I likely would have been able to create a playable version of Tripleopolis, working from the spec/concept, through the design, and finally to the development and testing, but it would have taken me a considerable amount of time – likely at least half a year (outside of work and family commitments). It’s also worth noting that my experience of creating games is very limited, and there are many disciplines involved that I only had an theoretical knowledge of at the start of this project.
As of today, Tripleopolis is a fully playable game, although it’s not perfect. While there are some graphics I’d like to improve and some game aspects I’d like to refine, the fact that I was able to leverage AI to create the first version of a game from concept to completion in a matter of months has completely blown me away!
In upcoming blog posts, I’ll explain my findings in terms of design, development, and overall thoughts on how AI will impact day-to-day tasks such as communications and project planning. However, I can already say that my concerns regarding AI replacing my capabilities have greatly diminished. Instead, the main takeaway (to this point) is just how much more productive AI can make people who are receptive and willing to learn how to use it – and from this standpoint, Tripleopolis has already proved what I needed it to.