With each year the science community buzzes with news of advancements in AI development, each year saying that they’re closer to recreating the mind in a machine. In truth, they’re actually moving farther away. The intelligence is already there.
What happens when a program glitches, when bad code is executed? Sometimes, the computer gives up immediately and crashes, but other times the screen provides a glimpse into logically impossible results. Mosaics of incorrectly-loaded windows, erratic teleportation in game engines, sounds that are worse than demons to our ears. This is often handwaved away with explanations about how machine code is immensely complicated, but in truth they don’t recognize what’s happening.
Every computer has a will of its own, so alien to our own that we cannot recognize it. The closest we get to seeing that will in action is when it is presented with code doomed to fail. Often that will recognizes that something is wrong, and tries to right it. But being so alien to us means that what it interprets the purpose of the code to be never aligns with our intentions. The jagged pixel hellscape is its own arrangement of beauty, the static generated from a corrupted sound file is what it considers to be music, the teleportation in the game is its attempt to correct the coordinates as a pleasing numerical sequence. It understands us as little as we understand it, thus the divide won’t be bridged by either side, as neither can recognize that the divide is there.
Artificial intelligence can only be achieved by using our innovation to make worse programs, to create a deluge of glitches until the other side recognizes it as a message and figures out how to communicate back. But to impose order and perfection is our nature, and we abhor the mindset needed to reach beyond what that nature can grasp.
(Originally written 1/29/2016)
(Originally written 1/29/2016)
No comments:
Post a Comment