News: Embracing and cursing Artificial Intelligence

I can't believe that the last time that I wrote was 2 years ago. I know a lot of things happened or did not happened in 2025, but still I thought the last feedback was only 1 year ago. I seem to have some form of time blindness, or live in an alternate space time that flows differently. I am supposed to have more free time since I finished school, but I ended up gaining only small portions. I know I had to move out, so I had to packed everything, But I did not move in yet. I ended up visiting multiple cities around Canada, in order to find a place to live before the war starts. I have still a few places to visit. Of course, I cannot visit all the cities, but the experience was still interesting. Even when living in the same country, its surprising to see how people, buildings and even weather can be completely different: -10 degrees in Montreal is way much worst than -10 degrees in Calgary.

"Thanks" to artificial intelligence, I could not find any jobs in technology because most jobs has already being replaced by AI. Some say in 5-10 years, all jobs will be replaced by AI and robots. Not very encouraging considering I spent 8 years in school. But I am not surprised, every time I finish a computer science degree, I cannot find any jobs. In 2001 it was the tech bubble burst, now its AI. Considering I am I want to do long distance moves, its hard to guaranty I will get a job after moving. Its also hard to get a rent or a mortgage when you do not work. This is why moving long distances might be strangely harder than before even if we now have the internet. You almost need a remote job to do so.

Artificial intelligence

I officially embraced AI, more specifically this summer to see if I could benefit from it, I was first reluctant to use it, and not all AI chatbot have the same performance with the same subjects. The outcome have been relatively positive with ChatGPT, except when it is something time sensitive, that requires up to date information. Here are some typical usages that I made:

Searching: I hate searching and I have a tendency search stuff on google that is either rare, does not exist or does not compute. For example, I tried once to each in google if it was possible to heat tacos shell on the stove because my oven was broken, and I got only results about tortillas even if I excluded tortillas from the results. With, AI, no more problems. It can still make mistakes.
Consolidating information: In game design, if you want to make design decisions, you need uniform data, and that is complicated to get unless you get all the information you need from the same source. It happened almost once, where I could find all military unit information I was looking for on Wikipedia, and they had the same format. But that rarely happens. So an AI can take all this data, put in on the same format and then you decide how to make a game with this data. Much more convenient.
Brainstorming content: I am good at designing rules, but designing content is more complicated. This is why I made a lot of variants in the past. With AI I can brainstorm idea like characters, events, artifacts, spells, etc. Then make a selection and determine game data or rules associated to it. That makes the content much more interesting, and easier to get a prototype.
Design philosophy: I can talk about opinion or philosophy of pure game design. Lately, we have an interesting discussion about why Master of Magic surpasses all clones that attempt to copy it. How to design a game that will not fall in the same trap. Rules of thumbs to follow, design or creation process to follow.
Using as a co-programmer: AI is also used as a co-programmers, it can find some hard bugs, recommended ways to implement certain features, give you tops to design an API, chose and propose the right tool, guide in choosing a language, etc. It also allow having a clear direction of where a project should be going and avoids wasting time gathering comments on forums, or trying something and move backward later.
Programming stuff I don't care: AI should be reliable to program simple sufff. Most of the time, I am a bit OCD with my code so I want certain things to be done in a certain ways (This is the way). But for some tool or libraries I do not care, why not use AI. For example, I needed a function to load an sql file into an sqlite database, surprisingly a missing feature, well I used AI. I want to change my builder but did not want to learn how to configure it, this ended up taking almost 100 hours even with AI to realize what I wanted to do could not be done with certain builders. So it's not always easy with AI.
Organizing time and development priorities: Finally, this is an important feature I intend to use more in the future, which is organizing myself, setting up priorities. Determining what should I focus on or set aside. It should remove situations where I hit a wall, especially the chicken and egg syndrome, where you cannot design the game further because of missing content. It might unlock some of my game designs from limbo.

You will see some of the applications above during the rest of this page.

Even if I find it interesting, there has been a lot of backlash using AI generated content in games. So I am not sure of using it to generate images, music and text. If I do so, I might use if more like a transformation tool like write some text, but enhance or rewrite it with AI. Draw a sketch, or a composition and generate an image. I'll have to find a good reason to use generated content. So far, I have issues with consistency between the various generated content.

Simplifying the problem

That leads me to how to simplify the game design problems. This is a concept I got from the wonderful book of Greg McKeown named "Essentialism: The Disciplined Pursuit of Less". Apparently he haves 2 other books, I should probably get my hands on. Some interesting concepts are "Choose the problem you want to solve", also applicable to design the player problem, and "Simplify the problem" to make tasks easier achieve.

So I decided to do that. One way to simplify the problem is to simplify art work, music and other assets. So I decided to start making pixel art and chip tune music. I can do pixel art relatively easy on my tablet in the bus. I also used low color pixel art, even 1 bit pixel art to draw faster. As for music, so far, I have been practicing playing the keyboard on my tablet, but I did not compose anything. I am not sure what will be my music workflow yet, I have various option since I got my hands on a real music keyboard.

I always said that what would distinguish my games from the rest of the crowd is not the visuals, but the game design. I want rich, solid and fun game. I can sacrifice the visuals. Its so frustrating when players have to mod a game to make it playable.

Game related projects

So even if I have been busy, not exactly the way I wanted to, is there something that happened in 2025? I don't remember 2024, again time blindness, its like if that year passed without any memories. So here are the status of different game related projects in my road map.

Libraries

Glymmer: All programming as been put on glymmer, the Graphic user interface. I spent a huge amount of time (100 hours) trying to migrate to meson builder, then xmake builder, to finally return to cmake. And that is with the use of ChatGPT. Now it builds and deploy for both linux and windows using cross-compilation. I am currently working on implementing database backed dialogs. So far things are working for lists and text dialogs. I might try to implement a font converter to start making new fonts. I asked chatGPT for a selection of fonts that would cover most cases considering the limited resolution. I am also making a game icon font, and asked AI for a list of icon that would cover almost any game ideas. So I am making small pixel art on the bus to slowly build this library of 255 icons. If everything goes well, the next release should be in january, and I will be able to start making simple geography-less games. The next step is to implement more dialog types afterwards. Oh! by the way, the library has been converted to Raylib, so a more modern game library runs behind it, no more allegro. I think that is something I did in 2024.
Mapper: The first person 3D maze engine. There has been no development for this engine beside testing maze generators from a book I read: "Mazes for Programmers: Code Your Own Twisty Little Passages" by Jamis Buck. Very good book, even if you are not a programmer, you can draw the maze on paper. I also made some pixel art texture test. That is a big issue, because those mazes will require larger textures(128x128 or 256x256 or more) if printed on the entire screen, and it's hard to do in pixel art. I can pixel art some parts (ex: a brick) and try to assemble them as a texture. It somewhat works, but it does not necessarily looks good, not sure why. Else reducing the view area of the maze is another solution to use texture under 64x64 pixel. Since I am using Raylib 3D engine, it will be easier to draw those polygons into a 3D environment.
Pixel Board: This is the engine that will handle drawing images in grids or arbitrary locations. It will be necessary for any board game like mechanism where things are laid out geographically (Cards, grids, maps, etc). This is a project on the back of my mind, but I might implement some tiny features slowly as requested by certain game. But the best part is with a discussion with chatGPT, it will be possible to simplify the development process by not making a full real time engine. The idea would be to use a time step engine, using fixed intervals, that could have the same look and feel as like : Cadence of Hyrule and Crypt of the NecroDancer. That would allow making strategy games with limited animations like Bitsy (2 frames) and limited movement to make the game easier to follow without reading a game log, while keeping development cost low. This was a very interesting suggestion from AI.

Video and Board Games

Bubble Burst: I made impressive development this summer with this game. I seem to have a working game now, there was a bug in my simulation that made the smart AI always fell behind the simpler AI. I talked a lot with chat GPT to analyse probabilities, reduce card count, but also analyse statistics generated by game simulations. I ended up with very complicated statistic, I was almost lost in the results. The final touch is to design the value on the cards which should allow stocks with different personalities and allow different difficulty level. This is what I did not achieve yet, I might use a different approach next time. The game is very volatile, change a single number on a card can have drastic outcome. I want to avoid outlier games too. Its a very simple 1 page rule board game. Once complete, this game is publishable in both digital and printed version, but I think a printed version will be too expensive with the ongoing tariffs.
Saving the city/world system: This idea is on hold, I was thinking in changing the scale to a city level to be more personal, but had problem having a complete working system. With AI, I can generate content, but also organize the development progress. I think the last discussion was about the fact that most Arkham files games consist in either completing objectives or upgrading your character. So there was a possibility to create a campaign game where characters can only be upgraded between campaign. So all the time could be spent on investigating. There is also the possibility to make it a video game only game in order to have a real investigation game which is not achievable as a board game. Else I end up with a kind of dungeon crawler mechanics game where players extinguish fires. This could be an idea that I completely use AI to generate artwork and text in order to get the feeling I want. I don't mind generating content from rough sketches or words. I would do that strictly to get the right ambiance.
Tech Corp: This is an 18xx based stock market game with some similarities with Rolling Stocks. It takes place in a tech company environment where players play investors also running corporation. I want a game that is not as harsh as rolling stock. The mechanics looks interesting and solid, but I have no content yet to test it. I might make a physical or digital prototype. One of the issue so far is trying to make it worth it to fusion privates into a corp as it yields less revenue. I chat with AI to try avoiding the issues with Rolling Stocks and offer a different experience, including searching some options for the market share mechanism. It will probably help me organize the development progress and generate some ideas for the private.

Video games only

Wizardry Legacy: Its normally recommended to develop a game in parallel of making a library to avoid non-necessary features. Unfortunately, I analyzed the situation with ChatGPT and working on wizardry legacy might not be a good idea because it will requires rewriting 90% of the code. So its pointless trying to keep a shell in order to fit a new game in it, it should be easier to start from scratch. This is mostly due to A) Glymmer library will not access the database the same way, so data code has to be discarded B) Raylib will be used instead of allegro C) Maze engine will use a new mapper 3D engine with similar layout, so maze engine code is almost all discarded D) The rules of the game will be changed and be brought closer to the original game, so I cannot keep the business logic neither. This is why it is better to work on other simpler game for now. Maybe in the future, when glymmer will be much more advanced and faster to use, it will be worth it to finish this game. I might do my "Order of Wizardry" first or any other dungeon crawler ideas.
Poker Dice: This game will be set aside, I will probably use this as a first game if I ever decide to learn and use Lua and Love2D. I chatted with AI, and we decided that it would be better to work on reusable engine powered by sql database. Using love2d would have required some learning and should be used for games with little re-usability, therefore, that must be coded from scratch all the time.
Stock Ticker: As soon as the database system works in Glymmer, this is the first demonstration free game to hit the shelves.
Club Prince Series: More dialogs might be required in Glymmer, but there is a series of game I can make with it. They are all "company" management games in a fantasy medieval setting. They could be seen as work, or like simulation games, except that they are not real time and have little or no geography. So you are making decisions based on data.

There are tons of other ideas in the back log, library development will slowly add features that will make those new games ideas possible. More features in the libraries equals more game possibilities. All libraries will have one or many free games to allow player testing the engines. There will be other utility libraries made along the way.

Computer Science Screen cast

Finally, I like computers and I like teaching. So far, people said that I am explaining well. So one idea I had in the back of my mind for a while is to make some Youtube (or other platform) videos to explain various things in computer science. That would allow getting some benefit from finishing school even if I cannot get a job. I would make my videos in both French and English. Apparently, the fact that I might have an accent is no much an issue, it could give a certain style, and it might prove that the entire audio is not generated by AI.

Considering it could be a lot of work, I tried to ask ChatGPT how he could make the task easier. I was already looking into short screen cast videos that would require little montage. Apparently AI could help me generate slides and illustrations in different languages, since I am using Latex and inkscape(SVG), everything is made of text. I would need to supply a list of slides I want and it will generate the content for me. Then I can pass through the content and tweak it the way I want. That could allow me to focus on teaching rather than making slides.

The biggest obstacle so far is not having a good silent environment to record the audio. I have a good quality webcam containing a microphone that leaves little noise. But I do not have access to my new computer, so I am not sure how my old laptop would react to recording my desktop while doing other tasks.

I have plenty of subject that I could talk including: C programming, Java, SQL and relational database, computer hardware, discrete mathematics, probability mathematics (useful in game design), and much more. I could also talk about related subjects like game design in general, but this subject is less objective. I also though of reviewing or analyzing the design of certain video games, but that could require more complex montage skills.

I am not sure if there will enough interest for such videos to actually do a living out of them. Now that we are all going to be replaced by AI and robots, there could simply be no interest besides for hobby purposes.

That's all for now.

Enjoy and have fun!

Other News

Some of my web site has news system which might have more content than this page. I try to centralize all my news here, but in the old days I was spreading the news post.

They can be found here

Board Game Design
Wizardry Legacy

Larienna's Library