= 27d567bc2d48d9f40e9ccc20ea370b15)
morganism comments on [LINK] Concrete problems in AI safety - Less Wrong
You are viewing a comment permalink. View the original post to see all comments and the full post content.
= 783df68a0f980790206b9ea87794c5b6)
Loading…= b715d02e85f7b3b4ae65ca3d3e450868)
Subscribe to RSS Feed
Powered by Reddit
= f037147d6e6c911a85753b9abdedda8d)
You are viewing a comment permalink. View the original post to see all comments and the full post content.
Comments (6)
these folks say that you won't be able to sandbox a AGI, due to the nature of computing itself.
Assuming that a superintelligence will contain a program that includes all the programs that can be executed by a universal Turing machine on input potentially as complex as the state of the world, strict containment requires simulations of such a program, something theoretically (and practically) infeasible.
http://arxiv.org/abs/1607.00913v1
But perhaps we could fool it, by poisoning some crucial databases it uses in subtle ways.
DeepFool: a simple and accurate method to fool deep neural networks
http://arxiv.org/abs/1511.04599v3
Sandboxing just requires that you be sure that the sandboxed entity can't send bits outside the system (except on some defined channel, maybe), which is perfectly feasible.
Citation needed.
In software, it's trivial: create a subroutine with only a very specific output, include the entity inside it. Some precautions are then needed to prevent the entity from hacking out through hardware weaknesses, but that should be doable (using isolation in faraday cage if needed).