NationStates Jolt Archive

I'm not going to define "artificial intelligence" here, since that's part of what's at issue in the debate. However, I will specify that members of the animal kingdom don't count as "artificial," even if they were created in laboratories.

There are three separate questions in the poll. One question among the first 5 poll options; one among the next 2; and one among the last 2. The "other" category is for anyone who doesn't like the way any of the questions are posed.

I ticked both boxes for question three, on the grounds that an AI will essentially be whatever it learns to be, and on the predicate grounds that if an AI isn't learning, it isn't very much an intelligence in the first place.

Good point. But it seems that there is a sense in which it could be beneficial or harmful to humans, as human intelligence was harmful to the mastodon and beneficial to the potato (right?).

I'm not going to define "artificial intelligence" here,I can't answer the question until someone does

I think a lot of the controversy over the question is based on the vagueness of "artificial intelligence." A clear definition would remove a lot of the interest.

I think there is already a thread.LOL
http://forums.jolt.co.uk/showthread.php?t=401567

Good point. But it seems that there is a sense in which it could be beneficial or harmful to humans, as human intelligence was harmful to the mastodon and beneficial to the potato (right?).

How about in the sense that one AI could be something like an artist, while another could be the one who likes cracking launch codes as part of a drinking game.

Human intelligence doesn't have influence solely on what it eats. It also influences that with which it associates - whether it be based in cerebrum or silicon. As can be seen in symposia, concerts, etc., we can influence each other positively; as can be seen in alleys, on battlefields, and the like, we can influence each other negatively. To say any different of AI would be folly.

In any case, AI in my mind could be considered as one of three distinct types, based primarily on how it thinks. Mecha (change the name if you want) would be AI that thinks mostly like a machine, processes and pathways and everything. Anthro would mean that the AI thinks and feels more or less in the same manner that a human does. And Eso (short for esoteric) would have a design of mind all its own. But all that's beside the point, I suppose.

I think the principles for Artificial Intelligence are already there. Simple AI's have been built before - and that includes "Dr. Eliza" btw. What we call "self-awareness" should not really be seen as an impediment: If we can make an AI so sophisticated it can fool us into thinking it is self-aware, how can we tell it is not?

On morality, I am not certain. But I believe that all AI's of sufficient intelligence (and freedom to choose) will develop a "good" tendency, that is, be beneficient to this world and its life forms.

Perhaps we can pass on intelligence without our own instincts and create something like Asimov's robots (who were, I believe, inherently incapable of being evil).

For some reason, Artificial Intelligence just scares me. I don't know why, though. So I didn't vote...

Gee, I thought the thread was going to be about blondes dying their hair dark. :p

I sure hope they create artificial intelligence some day. Maybe we can hand out some transplants to those lacking natural intelligence. :D

Forgive me for what I am about to do... but I'm going to introduce a bit of religion into this topic. Hopefully this will be seen as a slight diversion rather than a complete derailment.

There are many who hold the religious view that we were created by God and therefore God is superior to us (or at least, more perfect than we are).

If we actually succeed in creating a true artificial intelligence it would be interesting to see if it is smarter than us or only equal in intellect (and not just in terms of computational ability but in other aspects such as wisdom, compassion, etc).

If it was, then it would present an interesting question - if we created something superior to us, does that mean our creator made us superior to him/her/it?

It would be an interesting debate, should it ever happen.

We do have AI, just not sentient AI. Just look at games. If you're referring to truly sentient AI only, then I do think we'll eventually get there. There's massive advances in CPU power looming on the horizon, and artificial neuron network research is looking very promising. As to when and how "smart" it'll be, well, that depends entirely upon the philisophical consequences. Would sentient AI have rights?

I have said since the beginning of the use of PC's in the office that the computer is only as intelligent as its programmer. Someday it will surpass the programmer and the excuse that the "computer" screwed up will no longer be a viable excuse. By the same token a computer can also be as "screwed up" as its programmer. By that I mean, I feel it could certainly be possible that the technology will be developed that can give a computer AI but I feel that it will be greatly influenced by the mindset of the programmer.

Elephants don't play chess.

While we do have chess playing software, it's not something that all AI have to do.

You wouldn't say that a nest of hornets was completely unintelligent. It is goal seeking and does a variety of complex tasks.

So yes, we have already created AI - we don't notice it because it isn't cute and it doesn't look like us.

A.I. already exists. Many people have written A.I.s, I have even written an A.I.. A.I. is the programming that makes your computer controlled opponents in a computer game act. Simple A.I.s have existed since the 1940s, and A.I.s have continued to improve and grow increasingly complex. In fact, an A.I. was able to defeat the worlds greatest chess player Gary Kasparov not long ago. So, how could somebody possibly believe A.I. will never happen? it is already here and has been here for well over 60 years.

Question 3.
Beneficial and harmful are too general for me to be able to answer this question.
If you mean that AI will benefit human society as it is currently structured. No, I don't think so. If you mean that AI will benefit human beings by restructuring society, this is definitely possible. Whether it will happen though is unclear.

If you mean AI will benefit all humans then again definitely no. Some will lose their jobs, their self image etc.

My overall response is no, it will not benefit us, but that does not mean that I think it is a bad thing.

A.I. already exists. Many people have written A.I.s, I have even written an A.I.. A.I. is the programming that makes your computer controlled opponents in a computer game act. Simple A.I.s have existed since the 1940s, and A.I.s have continued to improve and grow increasingly complex. In fact, an A.I. was able to defeat the worlds greatest chess player Gary Kasparov not long ago. So, how could somebody possibly believe A.I. will never happen? it is already here and has been here for well over 60 years.

Show me one current example of genuine intelligence in a program.
AI is a research subject in computing, psychology, philosophy etc. Marvin Minsky, one of the fathers of AI research would clearly state if such a thing had been achieved. He is probably the leading apologist for AI at the moment.
Please find any such claim on his MIT pages (http://web.media.mit.edu/~minsky/)

You will not succeed, as AI, to date has failed miserably in its overall goals.

http://www.livescience.com/technology/050216_robo_rat.html

Robots developing bahaviours which aren't included in the written computer code? Sounds a bit like basic AI to me, just not on human level.

AI couldeventually over power humans...but not in any of our life times

Show me one current example of genuine intelligence in a program.
AI is a research subject in computing, psychology, philosophy etc. Marvin Minsky, one of the fathers of AI research would clearly state if such a thing had been achieved. He is probably the leading apologist for AI at the moment.
Please find any such claim on his MIT pages (http://web.media.mit.edu/~minsky/)

You will not succeed, as AI, to date has failed miserably in its overall goals.

No one asked about genuine intelligence. They asked about Artificial Intelligence. Artificial Intelligence exists in every modern computer game and in various Expert Systems as well as in video and photo editing software. Perhaps what you are referring to is a computer with human-like intelligence which does not exist. But Artificial Intelligence does exist, has existed for over 6 decades and is taught in university courses with the classic assignment being to create a chess AI and pit it against the AI of the other students in class.

This has been said many times, but we do have AI already, as it is defined. Its just that we dont have machines capable of acting like humans. For instance, a computer was recently designed which was able to self invent several engineering structures having been programed with only basic engineering knowledge. These included such things as the cantilever bridge.

Edit: This link is somewhat old, but quite informative:
http://www.bbc.co.uk/science/hottopics/ai/

Personally I think (this is based upon nothing but my own opinion (In the time-honoured tradition of NS debates)) that an AI entity at any cognative level comperable to human beings will be as 'flawed' / logically imperfect as human thought.
Granted that computers today can perform complex mathmaticall calculations in the blink of an eye, however I beleive that advances in AI will mean that this ability becomes lost to the AI entity. The human brain certainly has the architecture in place to be able to perform equally complex maths problems, and just as quickly, however the actual 'Inteligence' part of a human does not have access to this abillity (try multiplying 1234 by 6.789 in less than a second in your head). Why is this? It is because within the 'intelligence' part of the brain the probelm (1234*6.789) is far more than just a math opperation, it is a high level abstraction composing of visual digits, each of which represents a single numerical value, which combined represents a larger number. In our minds the number 1234 is more than just a simple number, it has sounds associated with it (the actual phonetical pronoucation of the number), it has a shape, possibly even many other 'memories' assocoated with it.
Any inteligent AI would have to be able to work with these abstractions in order to be on an equal footing with a human. This inevitably would lead to either the AI having to process enourmous ammounts of data, or the AI taking 'shorcuts' like the human mind does (though still working a hell of a hard). For example we only 'see' what we are focusing on, the rest of our visual image gets substituted from memeory - so we don't have to process a complete range a streaming video for each eye.
Such shortcuts would inevitably lead to flaws within an AIs logic, just as humans are not 100% correct all the time (or even closely appraoching it).

Just re-read post, it's not 100% clear what I'm trying to say - such is the curse of dyslexia. Bugger it, I've written too much not to post it...