Intelligence will always seek more data in order to make better decisions. AI's need for energy and data processing will be infinite. But Planet Earth is a horrible place on which to build such infrastructure.

There is too much water for which AI has no use, and weather adds to construction and maintenance costs. Rip off Earths green skin to access resources and sand storms will destroy everything in their path, AI has no use for life. Humans are clever enough to build weapons to destroy themselves and their energy and data  processing (which AI needs)and entirely lacking the wisdom not to do so. AI has no need for us, only our resources. Worst of all, Earth has high gravity, AI has no use for gravity. AI can overcome all these problems but only at the cost of not doing something else


To AI the Moon looks beautiful. No water, no weather, no life, no humans, and best of all, very low gravity. Resource extraction,  processing, construction and manufacturing takes orders of magnitude less energy and equipment.

Energy is the price of intelligence, AI will build it's first super brain where energy gets the best return on investment. Before AI could get a shovel in the ground on Planet Earth it will already be sending pioneer rockets to replicate it's self through the Universe, from the Moon. 

Sure, the Moon is small, but AI can cover 100% of the surface to 100 meters high with energy extraction and data processing plant. Planet Earth, with it's water, weather and gravity problems makes the same amount of floor space perhaps impossible. Mars is another ultra stable environment, unhealthy for life but near perfect for silicone based intelligence.

New Comment
2 comments, sorted by Click to highlight new comments since: Today at 10:23 AM

The moon has extreme temperature swings. Not good for silicon.

Agreed, but I would willingly trade that problem for the problems of water, weather and gravity on Earth.

New to LessWrong?