This is conceived as an informal and spontaneous annex to my more extensive blog, Grand Strategy: The View from Oregon

Interested in future Studies? Consider contributing your expertise to Project Astrolabe, investigating existential risks to earth-originating life, intelligence, and civilization.

23rd January 2013

Post with 1 note

An Idea for the Employment of “Friendly” AI

What sort of unlikely things might “friendly” AI do for us?

First off, what is “‘friendly’ AI” (or FAI)? Wikipedia defines it as follows:

A Friendly Artificial Intelligence or FAI is an artificial intelligence (AI) that has a positive rather than negative effect on humanity.

In contradistinction to friendly AI we might want to also define hostile AI (HAI) — I don’t know what term others are using for this, but I’m sure that some term is being used to describe the possibility, since it is perhaps more prevalent in fiction than FAI.

The mentally ill AI in the classic film 2001: A Space Odyssey, HAL 900, began as FAI and morphed into HAI. This suggests that we ought to bring a developmental perspective to AI, since there might also be AIs that begin as HAI and morph in to FAI (something like this is the scenario in James P. Hogan’s book, The Two Faces of Tomorrow).

Superintelligent machines could obviously do a lot of things for us if they are “friendly” and sympathetic to human interests. The old futurist dream of an economy of maximized abundance, with most labor automated, and the great question facing civilization being what people will do with their leisure time, may yet come to pass.

The big and obvious benefits of FAI such as maximized abundance are relatively low-hanging fruit, conceptually speaking. What other benefits, somewhat less likely, might be expected from friendly AI?

Here’s one idea: think of all the thousands (if not millions) of poems, books, and stories that have been written and never turned into a film. One might have a superintelligent friendly AI review a book and custom create a film based on the story.

If you were dissatisfied with the result, or even wanted to change the original story, you could just tell your friendly, neighborhood AI, and have it make the appropriate changes. Maybe you’d like to have yourself written into the story. That could be done too.

The possibilities are limited only by the imagination.

Tagged: AIartificial intelligencefriendly AIsuperintelligenceFAIhostile AIHAL 9000


  1. geopolicraticus posted this