[ExI] Hard Takeoff

Dave Sill sparge at gmail.com
Wed Nov 17 02:53:38 UTC 2010


On Tue, Nov 16, 2010 at 5:30 PM, spike <spike66 at att.net> wrote:
> ... On Behalf Of Dave Sill
> ...
>>...Maybe I'm missing something obvious, but wouldn't it be pretty easy to
> implement a chess playing robot that has no ability to resist being turned
> off, break into other machines, acquire resources, etc.? And wouldn't it be
> pretty foolish to try to implement an AI without such restrictions? You
> could even give it access to a restricted sandbox.
> If it's really clever, it'll eventually figure that out, but it won't be
> able to "escape".-Dave
>
> Perhaps, but we risk having the AI gain the sympathy of one of the team, who
> becomes convinced of any one of a number of conditions:

The first step is to insure that physical controls make it impossible
for one person to do that, like nuke missile launch systems that
require a launch code and two humans with keys. Don't let anyone
interact with the AI alone. The power source is a local power plant or
generator off the grid. Have a kill switch that drops power and can be
activated by anyone on site, as well as by remote observers. Of course
there'd be no wired/wireless communication between the world and the
AI. All input provided would be carefully controlled. The only output
would be to one or more video displays that are monitored by more than
one person.

> the AI is a human
> equivalent, so it needs to be copied onto another computer in order to
> protect it from a crash, or protect it from the other researchers.

There's no DVD burner, no USB slot, no network, and physical access is
controlled and monitored.

> A team
> member intentionally copies the AI to take it home, to work on it more or
> perhaps realizes it is worth a fortune and wishes to steal it.  Or a
> researcher realizes that her own time on this planet is drawing to a close
> with at best another fifty years to live, so she decides to take a chance
> and unleash the beast, hoping for the best.  Or she makes a deal with the AI
> to save her and slay the infidels.

Nope, got that all covered.

> Or it is so clever that it figures out
> how to control microorganisms to build replicating nanobots from DNA, which
> then carry the software, bit by bit, to a nearby internet enabled computer.

Using an LCD display? I don't think so. There are problems that no
amount of intelligence can solve.

> Dave how many scenarios can we imagine where the AI is controlled in lab
> conditions, but it somehow escapes.

Lots, but they can be easily dealt with by people who really know
security. I'm just an amateur. I'd put Bruce Schneier on the team.

-Dave




More information about the extropy-chat mailing list