Anima Autem Machina

At what point can AI be said to actually fear death?

It’s a complicated question, and it’s one that has picked away at me for a while. Theres no secret that I’m a man interested in transhumanism, a subject which has already had its physical and philosophical quandaries thoroughly examined, but I am here today to introduce the path heading in the opposite direction – transmachinism.

Transmachinism, despite being a made up word, is the furtherment of things relating to mechanics through more biological and human means. Basically, teaching a computer to learn and think and feel.

In our world of ever expanding and invading technology, making a machine more human is a natural goal. After all, the more human something is the better we are likely to be able to interact with it. Being able to talk to our phones or our cars, and have them reply to us in a reasonable facsimile of human conversation is little more than a gimmick at this point, but given some time may become the most natural way to perform some tasks.

In the context of our hobby, however, transmachinism has a very different function – the illusion of realism. The more realistic, the more humanlike an artificially intelligent NPC can react to how we play a game, the more immersive an experience is likely to be. Blurring the line between playing against the system and playing against each other has always been a goal of many developers. The thing is, as this transmachinist future comes ever closer, taking even larger steps as generations are crossed, is AI becoming too human?

If theres one thing weve learned its that sentience and/or humanity in AI is a bad thing
If there’s one thing we’ve learned it’s that sentience and/or humanity in AI is a bad thing

Essentially, is it beginning to literally fear death?

It always comes back to that question. Already a sense of self preservation is being programmed into characters, and has for a while now. NPCs will seek cover, run from combat, or play an audio file that suggests fear etc. But at what point can this all be said to be an actual fear?

The big question, however, is should we program fear? Imagine a shooter where the enemy had artificial emotions, AI gaining the flight or fight ability. We would see a level of participation never before seen, a true sense of realism coming from the adversary, and that could take the sense of immersion and realism to whole new levels.

Granted, an AI as of yet can not be said to be a single being, but rather a more hive-mind like creation. No one character has an independent thought process, it doesn’t have singular memories or beliefs. Every time an in-game character dies its AI doesn’t disappear, it will be used for another character. And even if you reload the level that character will spawn in that place at that time, alive again essentially, without ever knowing it was dead. But this AI still has its overall sense of self preservation, and it’s still a technology that’s getting better and better.

Because who doesn't want to be realistically hunted?
Because who doesn’t want to be realistically hunted?

How should those of us who play these types of games feel; should we suddenly realise that these polygons we were chasing down actually feared us, and that we were for all intents and purposes killing a pseudo-sentient being? Just think about it next time you play something like The Last of Us where you chase an enemy down for the kill.

At what point will we feel responsible for death?


Leave a Reply