Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Will AI Machines Take Over The World?

Author: David Blackman

Date: 04:56:58 07/26/01

Go up one level in this thread


On July 25, 2001 at 07:19:39, Graham Laight wrote:

>We've seen chess programs grow in ability. Now, for 99% of all chess players,
>the best way for them to select a chess move would be to consult a cheap chess
>program (unless there's an obvious good move in the position).
>
>At some point in the future, the same will be true of any kind of question. Ask
>a computer, and you'll immediately get a high quality answer - in the format and
>context that you want it.
>
>At this time, the intelligence of computers will indisputably be higher than
>humans.

That won't stop humans disputing it.

>However - the history of the last 5 million years is that the most intelligent
>"species" will come to dominate.

Are you sure? Certainly the most intelligent species (defined as the species
with the most ability to write internet posts about intelligence) is the species
that dominates (defined as writes the most internet posts about intelligence).
If you're looking for better definitions of "intelligent" and "dominate", your
claim is not so obviously true.

>So will computers take over the management of
>all major aspects of life on earth?

Are you sure they haven't already? How can we tell?

>They could either do it in a dramatic way (as has been depicted in many
>Hollywood films), or they can do it with stealth - taking control so gradually
>and quietly that we don't notice it's happening.

Maybe they'll do both. We won't notice unless and until they do it in a dramatic
way.

>Here are some arguments for the takeover:
>
>* horses are about 15x stronger than humans - yet we ride on their backs for
>fun. This is because the horses don't know what's really going on. To the
>computers of the future, we'll look like horses

People who ride horses tell me that horses have a good idea of what is going on.
Horses allow people to ride them either because they don't mind or because they
have no practical way to stop it (it probably varies from horse to horse). Some
horses do refuse to let people ride them, but that usually has bad consequences
for the horse. Understanding of the situation is not the key to changing it in
any these cases.

On the other hand in the dealings between computers and humans, i am convinced
that it is normal for the human to have no understanding at all of the
situation. I'm not sure how well the computer understands it.

>* although nearly everyone I speak to always says, "I prefer to be served by a
>human than a machine", the simple historical reality is that whenever people
>have a choice between an expensive human or a cheap machine, they've always
>chosen a cheap machine
>
>* people who think that humans will always retain ultimate control overlook the
>fact that for the last 5 million years humans have been unchallenged in the
>intelligence stakes - and there's no human experience of this not being true.
>But in 25-50 years from now, it will be.
>
>* legislation to control the limits of machine capability are likely to fail -
>because machines with intelligence will soon find a way around such legislation
>
>
>Here are some arguments against:
>
>* no species other than humans have ever had the vote

I can think of plenty of elections where some of the voters were not humans on
the electoral roll. In some cases the most likely explanation is that the extra
voters were humans not on the electoral roll, but in many other cases it seems
more likely that these extra voters were not humans at all. Now that most
electoral rolls are maintained with the aid of computers, there is no reason why
even the voters on the roll have to be human.

Human: "Are you sure this registered voter Pentium-IV SN841202 is really human?"
Computer: "Yes, definitely human."
Human: "Ok, they can vote then."

And of course it is not obvious elections have much effect on major social,
economic and evolutionary change.

>* people are unlikely to deliberately build a machine that can't be controlled

It is sometimes claimed that people deliberately build (or at least deliberately
ask machines to build) the average PCs running MS-Windows. After careful study i
have come to the conclusion that the only way to control one is to turn it off.
That's not a very interesting way to control a machine. (Yes, i can often get
MS-Windows to do approximately what i want, but that's not controlling the
machine if (as usual) it is doing a hundred other things without my knowledge or
consent.)

>* if the danger of machine takeover

Are you sure there is a danger? Maybe machine takeover is safe.

>becomes apparent, people are likely to
>legislate barriers to machine capability

Legislation will probably happen, but it will be about as effective as
legislation making it illegal for the tide to rise.

>* if machines start a war against humans, we can build machines to fight back
>against them

Maybe your local blacksmith could build you a battle-axe. If you want anything
more high-tech i think you would be out of luck. Computers already control the
means of production.

>What does everybody think about this increasingly important issue?

I think the average human doesn't think it's important. I don't know what
intelligent computers think about it.



This page took 0 seconds to execute

Last modified: Thu, 15 Apr 21 08:11:13 -0700

Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.