9
$\begingroup$

In Chapter 26 of the book Artificial Intelligence: A Modern Approach (3rd edition), the textbook discusses "technological singularity". It quotes I.J. Good, who wrote in 1965:

Let an ultra-intelligent machine be defined as a machine that can far surpass all the intellectual activities of any man however clever. Since the design of machines is one of these intellectual activities, an ultra-intelligent machine could design even better machines; there would then unquestionably be an "intelligence explosion," and the intelligence of man would be left far behind. Thus the first ultra-intelligent machine is the last invention that man need ever make, provided that the machine is docile enough to tell us how to keep it under control.

Later on in the textbook, you have this question:

26.7 - I. J. Good claims that intelligence is the most important quality, and that building ultr- intelligent machines will change everything. A sentient cheetah counters that "Actually speed is more important; if we could build ultrafast machines, that would change everything" and a sentient elephant claims "You're both wrong; what we need is ultrastrong machines," What do you think of these arguments?

It seems that the textbook question is an implicit argument against I.J. Good. Good may be treating intelligence as valuable, simply because man's strengths lies in that trait called "intelligence". But other traits could be equally valued instead (speed or strength) and sentient beings may speculate wildly about their preferred traits being "maximized" by some machine or another.

This makes me wonder whether a singularity could occur if we had built machines that were not maximizing intelligence, but instead maximizing some other trait (a machine that is always increasing its strength, or a machine that is always increasing its speed). These types of machines can be just as transformative - ultrafast machines may solve problems quickly due to "brute force", and ultrastrong machines can use its raw power for a variety of physical tasks. Perhaps a ultra-X machine can't build another ultra-X machine (as I.J. Good treated the design of machines as an intellectual activity), but a continually self-improving machine would still leave its creators far behind and force its creators to be dependent on it.

Are technological singularities limited to ultra-intelligences? Or technological singularities be caused by machines that are not "strong AI" but are still "ultra"-optimizers?

$\endgroup$
4
$\begingroup$

That would be a no for speed or strength, if you have a super strong entity but it cannot research new materials, it will be quickly limited, same thing for speed, Basically, you need something out of their field to improve them, which makes a runaway improvement impossible.

Though, we already have super strong and super fast machines, those are cars, trucks, hydraulic presses, industrial exoskeletons etc... But, even though we can build better ones through the use of the old ones, we still need to research stuff that can't be improved by old ones.

What we need for a singularity is a field where an improvement in it makes further improvement easier. And I don't know a field where this doesn't involve intelligence. If there is one, that may be possible to have a non intelligence driven singularity there.

$\endgroup$
0
$\begingroup$

"Monte Carlo" seems to be the best method currently for algorithmic creativity. (i.e. the machine makes random choices and sees if they lead to anything useful.)

While it appears obvious that creative connections formed out of understanding are superior to those which are random, if the machine is fast enough, it should be able to win out by pure "brute force".

i.e. Evolution, prior to human guidance, has not been been based on intelligence.* Rather, evolution has been based on random mutations that are either beneficial or detrimental.


*The caveat is that humans creating algorithms and altering genes (either in a lab or through animal husbandry and horticulture) can be said to comprise a new form of evolution that is actually rooted in human intelligence and desire.

$\endgroup$
-1
$\begingroup$

Well, that's obviously a huge debate that could be discussed for hours. I did Artificial Intelligence as a minor in my graduate studies & have spent a great deal researching futurists, including the all too well known Kurzweil...

In my opinion, the singularity does not require an emergent ultra-intelligence - humans can do it themselves (although, I think GOFAI could be bound to humanity and would be the far better option entering a singularity). Ultra-efficient utilities are the direction to go if AI becomes demonized ennough - essentially man and machine become one. The problem with this is that you don't have an objective agent ushering in the singularity for humanity, instead, you have selfish elites crossing a bunch of small singularities smeared across a period of time - what obligation would they have to come back for you...? A machine hardwired with a few lines of code to preserve humanity or forward humanity would be my choice.

$\endgroup$
-1
$\begingroup$

This May not be what you were looking for, but technically yes. Although not for Speed and Strength. But you could randomly guess new Mathematical/Physical/chemical solutions to become more efficient in random guessing (basically anything that allows the machine to compute faster and to maybe simulate the effect of those findings) thus technically achieving something similar to a singularity, without having to have any Intelligence at all actually (or just on a human Level), since you could just brute force all.

Is this efficient? No, not even close to being in any way feasible. Does it work? Technically, Yes.

It would be a singularity of sorts, since it improves itself continuously, but it wouldn‘t need to improve its own intelligence.

Of course, some findings might make it possible to become more intelligent, but let‘s just assume it doesn‘t apply those findings to itself.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.