An Aesthetic for Artificial Intelligence

Creating an interesting narrative for SubLight, one that takes place over many millennia, has a unique challenge: who is the player, and how do they live so long? Some games, like the Civilization series, simply cast the player as some immortal ruler. That’s fine for Civilization, where the story of the game emerges from the player’s actions, but I wanted a more hand crafted narrative for SubLight. I came up with three main options, all of which had their benefits and drawbacks.

The player is the will of the people

Leviathan, 1651

The concept of the player as a faceless representative for their ark was my first idea. This solved the problem of why the player never seems to die: they’re simply the next incarnation of whoever is running the ark. Ultimately, this created more problems than it solved. For example, how do you have character development between different societies across thousands of years? While I don’t think the problem is unsolvable, I’d actually love to make a game about that, I think it was outside the scope of SubLight’s narrative goals.

The player is cryogenically frozen

2001: A Space Odyssey, 1968

Another solution I considered was adding consistent and safe cryogenic freezing to the SubLight universe. With the ability to sleep away their voyage between stars, we could explain how you keep running into the same people over and over again. Unfortunately, this would destroy a huge chunk of SubLight’s uniqueness as a game about generation ships. If we’re going to add cryogenic freezing, we might as well add faster than light travel. This was not acceptable, and additionally, I wanted various encounters that dealt with the dangers of cryogenic freezing, rather than using it as a narrative crutch.

The player is an artificial intelligence

Alien, 1979

Early on I dismissed the idea of the player being an artificial intelligence. I figured SubLight’s universe would consist of near future tech, and intelligent AIs seemed a little too out there for my taste. However, the idea grew on me more and more, especially when I considered how fun it would be to “upgrade” your AI with the ability to use empathy, aggression, and deception in conversations with other arks. Additionally, the player would have to balance the goals of their computer AI with the needs of the population they’re protecting. That dichotomy could lead to really interesting stories and tough decisions for the player to make. For example, do they save a distressed probe with a lonely AI, but risk their population in the process? How will they fend off mutiny when they make poor decisions? Making the player an AI added new problems, but they were interesting problems, ones that would add depth to the narrative.

Who is human, and who is AI?

Now that I resolved to have AIs take control of interstellar arks in SubLight, I needed a way to distinguish them from the humans of SubLight’s universe. The avatars in SubLight are currently works in progress, but I wanted something that was “good enough” to get the point across. I decided to use silhouettes for humans and abstract geometric designs for AI.

I made these avatars in just a few minutes, but they’re enough to use while prototyping encounters. I’m excited to work more on the AI avatars, I think there’s a lot you can convey through shapes alone, especially when taking advantage of pareidolia. There can be emotional depth to certain shapes, like the friendly roundness of circles versus the hostile edges of triangles and chevrons.

For fun, I also added a very computer-y background to the AI, with scrolling text. It’s visually noisy at the moment, so I’ll have to play around with blurring and contrast to make it more subtle.

For those curious about conversations in SubLight, you can check out an example of one below.

Not sure if YouTube has any plans to process a 1080p version of that video, so I’ll have to warn you that the text is much clearer on an actual build.

This encounter will replace the silly informal one in the last build. I wanted to really drive home the fact that the player is an AI, so I put them through a diagnostic check answering arithmetic and ethical questions. I also took this opportunity to give the player a preview of the abilities they’ll unlock when adding empathy, aggression, and deception modules to their AI. These upgrades allow them to add emotion to their responses, unlocking new branches in the narrative.


Keeping Track of Where You’ve Been

I spent this week responding to feedback from the previous build. The biggest change is the removal of the transit lockout. As cool as it was, and as hard as it was to say goodbye to it, I believe cutting it was the right choice. It broke up the flow of gameplay far too much. Instead I’ve added a more subtle red ring that spins as you travel.

I also added lines to show where you’ve already traveled. It’s a small thing, but I think it already adds a lot of useful context to the navigation area.

A video of the transit history in action.

New options have been added to the preferences menu, including a way to increase the size of interface elements. This will be particularly useful for folks with 1920×1080 monitors, which are particularly common but didn’t show conversation text clearly.

In addition to all this I spent time combing through and fixing a lot of little bugs and off-by-one-frame issues. I really dislike seeing weird popping or other graphical glitches in games, it shows a certain lack of polish.

Next week will be spent adding in the initial encounters for the main story, as well as a new resource: metallics. Metallics will be used for repairing your ship, though likely not until a future update. If I have time I’d like to add the ability to turn on a resource map, that lets you see what are good systems to pick up resources at.

That’s all for now!