If AI gets too clever, it might calculate that human beings are counterproductive to its aims, and eliminate them.
"[People] assume that intelligence equals consciousness - a misconception that annoys many AI researchers." Because a machine doesn't need to be conscious to calculate that human beings are not aligned with its goals.
The article explores how we can ensure that computers are prevented from wiping out the human race when they get too smart - a big off switch (but the computer could gain control of it, or push it inadvertently) 2) tech the computer human values or 3) ensure the computer is also coded to e.g. maintain a steady number of people crossing the street in Tokyo, a surrogate for maintaining human population.
No comments:
Post a Comment