Elon Musk and Stephen Hawking on artificial intellegence.

[sukumvit boy]

Lots of media hype ,but still some interesting thoughts about the threat of artificial intelligence.
Interesting news also about the San Francisco AI group that calls itself 'Vicarious' that is attempting to build a program that mimics the human neocortex , the area of our brain responsible for sensory perception, spatial reasoning, conscious thought and language.
Vicarious claims ,"a new computational paradigm", they call "The Recursive Cortical Network" .
http://www.designntrend.com/articles...telligence.htm

http://www.cnbc.com/id/101892104#.

Damn! Help me out on this trish , but it looks like we passed the Turing Test on a side street!?

[broncofan]

Hopefully Trish comes back soon and you two can discuss this. I don't know enough about it to say anything but I will be reading. Might even be some good opportunities for science puns;.

[sukumvit boy]

Thanks ,broncofan ,I didn't know that Trish was away. I figured I'd just let this thread die a quiet death if no one was interested.

[sukumvit boy]

More about .possibly, redefining the Turing Test.
http://innovationinsights.wired.com/...g-means-think/
http://hplusmagazine.com/2014/07/28/19075/

[trish]

Can Machines Think?

Yes. There are already examples.

People are machines. People think. Therefore, some machines think.

Because animals are closely related (genealogically) to humans and exhibit behaviors associated with thought in human beings, we have reason to think that some animals think as well. Animals are machines. Some animals think. Therefore some machines think.

Can thinking machines be dangerous by virtue of the fact that they think?

We certainly are. We’re the biggest threat to all other life on this planet. To what to we owe are vast power of destruction? Our brains. To what do we owe our tendency to be destructive? Our brains. However, it’s far from obvious that all thinking machines have the destructive tendencies humans seem to display. We may have a nasty case of bad wiring.

But human’s are an example of natural intelligence (yeah right). Is artificial intelligence possible?

This universe doesn’t contain artificial bodies. If it’s in this universe, it’s natural. Bird nests, bee hives, beaver dams, Romeo and Juliet and smart phones are all natural product of naturally occurring living things. The real question is

Can human beings create thinking machines (in ways distinguishable from human reproduction)? Notice how dificult it is to even define artificial intelligence, thought etc.

We are very very far from having definitions of thought, intelligence and consciousness. We are equally far from having any real scientific or even philosophical understanding of these concepts. Can thought be carried out by purely digital processes, or is it in some way essentially analog? (Since the neurons of the brain are not synched to a clock like a computer the brain has_in this sense_an analog component. There are idealized hybrid machines that realize non-Turing-computable functions.) Is “thinking” reducible to mathematics alone, or does physics and chemistry necessary to an understanding of the general concept of “thought”? We don’t know. The answers do not (to me) seem to be immanent.

But building a thinking machine doesn’t necessarily require a complete understanding of thought. Some man-made electromagnetic devices were in use before Maxwell published his famous set of electrodynamic equations.
Indeed we already talk to machines that appeal to our tendency to anthropomorphize. There are machines that lure us into talking and thinking about them as if they were conscious agents; i.e. when using my GPS I often think, “She wants we to take the next exit, so I should shift lanes.”

Personally, I don’t think we’re going to be taken over by our SkyNet overlords anytime soon. On the other hand, I do think that at some future time man-made machines may have the capacity to think, want and act autonomously. I also suspect these capacities arise suddenly as technology crosses certain (now undetermined) thresholds (in the sense that Turing machines attained their maximum capacity to mimic all other computational devices when they become universal).

The real question is, “Would you take the red pill or the blue one?”

[sukumvit boy]

Right , Trish , I also find it difficult to identify any imminent risk of 'computers taking over the world'.
As I said , lots of media hype there , mostly surrounding the movie , " Transcendence" ,which I haven't seen. Also the new book , "Superintellegence : Paths, Dangers , Strategies" by Nick Bostrom , which looks like an interesting read.
Superintelligence: Paths, Dangers, Strategies: Nick Bostrom: 9780199678112: Amazon.com: Books@@AMEPARAM@@http://ecx.images-amazon.com/images/I/61uLf1NhflL.@@AMEPARAM@@61uLf1NhflL


Certainly , advances in AI are going to change the way we live and work in ways we need to stare thinking about.
More interesting to me , and as you have expressed an interest in on other posts , are questions surrounding Machine Learning , Theory of Mind and how our brain has changed in ways that made us, as anthropologist Ian Tattersall says ," Masters of the Planet".

[trish]

When electronic computing machines made their first appearance on Earth it was said that machine language translation was just around the corner. The problem of translating text or speech in one language to another proved far more difficult than initially anticipated. A half century later and we still don’t have universal translators. We do have Google Translate, which is moderately successful at providing translations of text. Instead of bringing any true understanding to the task of translation, the Google program utilizes a combination of brute force with crowd sourcing. It breaks your text into snippets and searches the web for occurrences of those snippets in pairs of already translated documents; i.e. it uses computational speed (brute force) to search the web for prior translations of fragments of your text and then sews those translations together. Then it asks you how successful you think the translation is.

There are also chess playing programs regularly that beat our best chess masters. But these programs do not employ strategy, at least not as the term relates to chess. Instead they utilize their capacity to search the game tree at high speed. By sheer brute force the machine can look further down the tree of consequences than any human can and pick the best move.

These successes, though impressive, are not encouraging. It’s difficult to see how these sorts of successes further advance our understanding of linguistic meaning, thought, intelligent problem solving or consciousness.

Thanks to brain scanning technology we are making some elementary strides into the understanding of the human brain; this science is in its infancy and (I believe) it will be some time before we have an understanding of human thought and consciousness, if ever.

Unfortunately there is a lot of room for abuse. There is talk of using brain scans as lie detectors, something (I think) should be vigorously questioned. Machines that can outperform humans (even though they can’t think or understand) are already displacing us in factories. This shouldn’t be mischaracterized as machines taking human jobs; rather it’s humans making a profit by utilizing machines to do their labor instead of hiring people. I can hardly blame factory managers for the practice. After all, I’m reasonably well off (middle class) but I don’t hire a maid, instead I have a washing machine and dryer. I could even buy a dishwasher and a Roomba if I wanted. Intelligent or not, technology has profound influences: just look at our modern communication system and how it has changed our lives in just the last couple of decades.