Dan Dias

| Blog and Portfolio
Apr 22 2017

Source Control

The source for this post comes from an article on MIT Technology Review. I’ve always had a fascination with Artificial Intelligence for video game purposes; which isn’t true AI. More like ½ part loose decision making algorthims, 2 parts smoke, 2 part mirrors, and the last ½ suspension of disbelief. Anyway, check out the article if you want but I’ll summarize down below.

The Short, Short Version

The basic premise is that deep learning neural networks make their own inputs and outputs between parts of their network. In essence, they program themselves. This means engineers that create the networks aren’t able to easily understand the path of the input because they didn’t write it. The decision workflow ends up being so complex with larger and larger networks. People are working on ways to get AI to explain itself and give a human digestible reason for a decision. This is all a gross oversimplification of the article, but it’s the part that’s relevant to what I’m thinking about.

My Relationship With AI

There’s an ongoing fictional universe in my head that has evolved over time. It’s a story I’ve wanted to tell, but it’s never been ready. One thing has been key in the story for the years I’ve been thinking about it; true AI and its relationship with the lives of people. The main character has always been someone who could basically do magic by interfacing with their AI better than most. It always makes sense to me since some people have an aptitude with technical skills, but it did not seem realistic for the reason to simply be because they are better at talking to an AI. Considering the article, maybe it’s not as far fetched as I thought.

When I was young and didn’t understand how game AI worked, I entertained the idea that the computer player was secretly a human, somewhere else, competing against me. There was a rational part of my brain that knew that wasn’t possible but I liked the fantasy of it. I may have been a lonely child. There are games that have toyed with the idea of bluring the line between Artificial and Human Intelligence.

Do Androids Deep Dream of Electric Sheep?

Some of the concepts in the article, extrapolated out to the absurd/Sci Fi, could be useful in determining the soul and motivations of AI. I’m experimenting with Unreal and exploring the way to create a Souls-like game. I’m transmuting my universe once again (actually, more like using a small section). It takes place in the city of Utopia, the world’s first fully Smart city. Things went very wrong with this city because, of course they did. Basic idea is that different AI were in charge of running the city, with on central core to coordinate them.

I want to create all different types of AI cores, basing them off mythological archetypes. The puppet master designed to bring entertainment to residents who embodies the Trickster. The core that dictates traffic and passage throughout the city who takes cues from the liminal deity, Hecate. It has been fun thinking of the different aspects that make a city run, applying them to mythological beings, and then considering their AI personality and motivations. What does their model of the world look like and how do they act on it?

Rocking Basscss