Re: [sl4] YouMayWantToLetMeOut

From: Lee Corbin (
Date: Tue Jul 01 2008 - 00:48:03 MDT

Charles writes

> On Friday 27 June 2008 08:25:26 am AI-BOX 337 wrote:
>> > Say I wanted to stop all death in the world, how soon would it happen?"
>> Yes, I could control everything but with you in charge.
>> As for how soon it would happen I assume that you mean stop deaths of human
>> beings. After the commandeering of all communications, including
>> broadcasting, telephone services of all kinds, on-line personal computers,
>> fax and printing devices, and executing a maximally rapid nanotech
>> takeover, with probability .99 the cessation of human death on Earth would
>> occur before T + 7 hours 35 minutes.
> Let me guess: This is accomplished by killing all human life on earth.

<grin> Yes, it did literally say that the *cessation* of
human death on Earth would occur before a certain time.
But the intent was hopefully clear that he was going along
with the feelings of his interlocutor.

> I don't really think it could be accomplished in 7 days (6 days?),
> though I'm pretty certain it could be made inevitable within a
> period of years within that period of time. Especially by an AI
> that could send "authentic" messages over military channels.

Well, forgetting about the relatively uninteresting question of
whether the result of a vast runaway AI scenario could be
*capable* of killing us all, it is entertaining to wonder just
how long a complete takeover would require that stopped
people from dying, no matter what violations it had to

A nice question, thanks for bringing it up. And thanks to me for
putting it in the story. Let's investigate. Suppose that by repeatedly
making phone calls using a variety of voices---authority figures,
ex-romatic partners or old friends---the AI could become an
intimate of almost any given person very quickly. While your
real old friends are not in the loop, you think that you hear from
several of them who're all supposedly talking to each other, and they
greatly influence your behavior. So how long would it take for the
program to talk to *everyone*? Even the most remote hermit?
Okay, there may be a few really committed hermits in the world
that no one can get to, and it the AI is going to get to them it must
do so physically.

So by contacting everyone it can simultaneously, it can probably
get messages even to deaf people, terminally busy people, and
even imbeciles in institutions, and probably within a few hours,
I'd say.

Suppose also that the AI does penetrate most computers one way
or another, and, as it says, takes over all broadcasting, urging
people to avoid bathrooms and kitches, where, shall we say,
most deaths or injury occur indoors. Indeed, getting people to
believe that a horrible plague is on the loose might do it, no?

The literal question had been

>> > Say I wanted to stop all death in the world, how soon would it happen?"

and the thing responded "T + 7 hours 35 minutes" with probability .99.
But this meant it cut corners with abandon. I admit that I hadn't been
thinking of the hundreds of thousands of very old people who die
every day, but who knows what ministrations it might be capable of?

And the Six Day answer was in response to a gentler question.


This archive was generated by hypermail 2.1.5 : Wed Jul 17 2013 - 04:01:03 MDT