Is Human-Level Artificial Intelligence Attainable? The Skeptic's Perspective

[PIT2]

From the skeptic site:

According to the roboticists and their fans, Moore’s Law will come to the rescue. The implication is that we have the programs and the data all ready to go, and all that’s holding us back is a lack of computing power. After all, as soon as computers got powerful enough, they were able to beat the world’s best human chess player, weren’t they? (Well, no — a great deal of additional programming and chess knowledge was also needed.)

Sad to say, even if we had unlimited computer power and storage, we wouldn’t know what to do with it. The programs aren’t ready to go, because there aren’t any programs.

Even if it were true that current robots or computers had attained insect-level intelligence, this wouldn’t indicate that human-level artificial intelligence is attainable. The number of neurons in an insect brain is about 10,000 and in a human cerebrum about 30,000,000,000. But if you put together 3,000,000 cockroaches (this seems to be the A.I. idea behind “swarms”), you get a large cockroach colony, not human-level intelligence. If you somehow managed to graft together 3,000,000 natural or artificial cockroach brains, the results certainly wouldn’t be anything like a human brain, and it is unlikely that it would be any more “intelligent” than the cockroach colony would be. Other species have brains as large as or larger than humans, and none of them display human-level intelligence — natural language, conceptualization, or the ability to reason abstractly.87 The notion that human- level intelligence is an “emergent property” of brains (or other systems) of a certain size or complexity is nothing but hopeful speculation.

After more than 50 years of pursuing human- level artificial intelligence, we have nothing but promises and failures. The quest has become a degenerating research program89 (or actually, an ever-increasing number of competing ones), pursuing an ever-increasing number of irrelevant activities as the original goal recedes ever further into the future — like the mirage it is.

http://www.skeptic.com/the_magazine/featured_articles/v12n02_AI_gone_awry.html

I didnt know the AI field was in such a bad shape, but then again i don't know anything about AI either. What are the opinions here, does skeptic magazine have a point?

 

[K.J.Healey]

I don't know if I buy that since it hasn't worked for the past 50 years we should stop. I don't know much about the advancements in AI but I DO know that in the last 50 years our ability to expand the input recorded by these computers has improved greatly. From picture/camera recognition to mood recognition due to facial features. Being able to efficiently recreate the senses of a human brain I believe is the very first step in developing AI.

Some good links about the changes in AI can be found on this site:
http://www.aaai.org/aitopics/html/overview.html

 

[neurocomp2003]

also got to remember about the education required to study AI and the funding available out there to do human AI studies

 

[setAI]

how ironic- a so-called 'skeptic' who embrasses the most insipid form of vitalism- that the biological human brain is imbued with magic that equivalently complex systems could never posess-

the field of AI is exponentiating- it is only the old naive top-down high church computationalism that has faded- bottom-up approaches are emerging and evolving quite rapidly-

and at the end of the day- the worst case scenario is that we will just have to bootstrap part/all of the brains functions into various computational/mechanical/biotech analogues- unless you believe in magic this will work

 

[Alkatran]

The article bothers me. The author seems stuck on a computer emulating a person, when AI has plenty of success in other areas. How does he think the Mars landers drive around? Remote control? The fact is: computer aren't people, you don't use them the same way. Why simulate a brain when all you need to do is recognize characters or faces? There are better approaches.

 

[CRGreathouse]

On one hand, AI has not lived up to its promises. On the other, it's not in as bad a shape as the article would suggest. We have many robots/AIs which easily surpass insects in intelligence. We have fairly advanced voice recognition and object detection systems. Automatic proof generators and arguably all-purpose symbolic solvers are further advances in AI.

 

[Milo Hobgoblin]

and at the end of the day- the worst case scenario is that we will just have to bootstrap part/all of the brains functions into various computational/mechanical/biotech analogues- unless you believe in magic this will work

but much of the brains lower level function IS bootstrap.. thatwouldnt be the worst case scenario. it would be the only case scenario.

Its learning how to USE those boostrap functions that humans develop as they age to 5 years old.

AI is completely possible.. it will take a very low level programming language with huge amounts of flexibility... immensely flexible variables (like comparing that stack capabilties of 16 bit ASM to 64 bit ASM.)

It will also require chips that have the ability to rewire themselves. Neural connectivity is not static and is the basis of learning. A hard wired chip would have to simulate that in software .. VERY slow.

 

[PIT2]

In another topic some people made it sound like my idea of intelligence was a stretch of imagination (i believe bacteria are intelligent), yet i suspect these people are actually the same ones that call robots that use a camera to avoid obstacles, examples of AI.

It will also require chips that have the ability to rewire themselves. Neural connectivity is not static and is the basis of learning. A hard wired chip would have to simulate that in software .. VERY slow.

I wouldn't forget to mention consciousness. Its my opinion that when(if) we figure out how to create a subjective experience in any configuration of matter, then that configuration of matter will start behaving intelligently. This i think is the basis of intelligence, and i suspect all other approaches are merely attempts to copy the behaviour that this basis produces.

Its like trying to create a fire by putting orange,yellow and blue powder into a transparant balloon, shining a flashlight at it, and then shaking it.

 

[Chronos]

It is very plausible we may create an artificial race imbued with AI. It seems to be a logical step forward in evolution. Perhaps biological entities will become obsolete in our distant future. A scary thought. Program them with a sense of morality, and who knows where that might lead. They might hold, and judge us by our word.