438

LESSWRONG
LW

437
Personal Blog

0

Military AI as a Convergent Goal of Self-Improving AI

by turchin
13th Nov 2017
1 min read
6

0

This is a linkpost for https://www.academia.edu/35130825/Military_AI_as_a_Convergent_Goal_of_Self-Improving_AI
Personal Blog

0

Military AI as a Convergent Goal of Self-Improving AI
4J Thomas Moros
0turchin
2Corguive
2morganism
0turchin
0entirelyuseless
New Comment
Email me replies to all comments
6 comments, sorted by
top scoring
Click to highlight new comments since: Today at 11:48 PM
[-]J Thomas Moros8y40

Not going to sign up with some random site. If you are the author, post a copy that doesn't require signup.

Reply
[-]turchin8y00

Thanks, it was not clear to me that it is not visible to non-members.

New link on google drive - also commenting is open. https://docs.google.com/document/d/15D71qhhY-ZsAY7syzZsr1lKopTODbdeXVPElaPaIqyA/edit

Reply
[-]Corguive8y20

thank you people very much for all the help and information. i'm going to check the https://musclegurus.com/ link more carefully but as for now, i appreciate a lot your help!

Reply
[-]morganism8y20

Seems to me at 3.4, the first stage is to bring down all competing nations electric grids, thereby slowing trade, transportation, and hampering retaliation. Think of non-lethal mines. The side effects would be instantaneous, and lethal. And the military could easily see it as a defensible and deniable action.

this would be trivial to anything hooked into the net, and has access to SCADA coding and grid diagrams.

It would also behoove the local "friendly" AI , to interupt communications in it's home base, to keep outside influences from retaliating against it's base. So you would likely get a comms blackout , except when the AI wanted to communicate it's orders. It would also make sense to set up new frequency hopping algo's on existing lines where that would be possible (i.e. not old POTS lines?)

Reply
[-]turchin8y00

This is our accepted chapter in the edited volume "AI Safety and Security" (Roman Yampolskiy, ed.), CRC Press. Forthcoming, 2018

Reply
[-]entirelyuseless8y00

People are weakly motivated because even though they do things, they notice that for some reason they don't have to do them, but could do something else. So they wonder what they should be doing. But there are basic things that they were doing all along because they evolved to do them. AIs won't have "things they were doing", and so they will have even weaker motivations than humans. They will notice that they can do "whatever they want" but they will have no idea what to want. This is kind of implied by what I wrote here: except that it is about human beings.

Reply
Moderation Log
More from turchin
View more
Curated and popular this week
6Comments