[ExI] ai in education

[Keith Henson]

On Sat, Mar 7, 2026 at 4:03 PM Adrian Tymes via extropy-chat
<extropy-chat at lists.extropy.org> wrote:
>
> On Sat, Mar 7, 2026, 2:50 PM <spike at rainier66.com> wrote:
>>
>> -----Original Message-----
>> From: extropy-chat <extropy-chat-bounces at lists.extropy.org> On Behalf Of Adrian Tymes via extropy-chat
>> Cc: Adrian Tymes <atymes at gmail.com>
>> Subject: Re: [ExI] ai in education
>>
>> On Sat, Mar 7, 2026 at 2:17 PM spike jones via extropy-chat <extropy-chat at lists.extropy.org> wrote:
>> >>... The same reasons the military distrusts Anthropic would cause me to distrust it: we can’t be sure it won’t turn on us.
>>
>> >...How can you be certain, to the degree you are requesting of AI, that a human-run military won't turn on us?
>>
>> We can't.  We use all available resources and technology to prevent it.  So far so good.
>>
>> We are buying AI.  We need complete control of it before we can trust it with our defenses, using all available resources and technology.
>
> If the military runs well enough without complete control of humans, why would it need complete control of AIs, as opposed to similar reviews and checks?
>
> There are tales, from before paperclip maximizes, of those who pursued "power" at any price, or any one goal to the detriment of all others.  They tended to get their goal only technically, in a way that robbed the point of the goal.

The evolutionary point of power is that those who obtained it
(virtually all men) in the Stone Age often did really well at
reproducing.  We share this with chimps.  Our detection of power is
attention, which selection has made as or more rewarding than
addictive drugs.  The detector can be tripped in situations where no
evolutionary advantage actually exists.  Climbing the power structure
in the Roman Catholic Church has not helped reproduction for a long
time.  Becoming a high-level scientologist isn't any more useful in
the evolutionary sense than being a heavy-duty addictive drug user..

> It may be that being intelligent (biologically or artificially), and being able to be completely controlled, are mutually exclusive.

I agree.  That's why Spike's model of what the military wants is flawed.

I am slightly reminded of the root-mean-square module I designed for
the military back in the early 70s.  When they were integrating it
into sonobuoys, they asked if Burr-Brown could change the response
time constant and make it symetrical.  The constant was set by the
math that generated the RMS output.  BB told them no, it was not
possible, if they wanted RMS, that's the way it is.

Keith
> _______________________________________________
> extropy-chat mailing list
> extropy-chat at lists.extropy.org
> http://lists.extropy.org/mailman/listinfo.cgi/extropy-chat