This is a bit of a long read, but an interesting article on how the most successful / powerful artificial intelligence programs out there are self taught and due to the vast amounts of data that they compile and consume to come to their conclusions, no one really understands how they make the decisions they do.   Yet they are very accurate.

https://www.technologyreview.com/s/604087/the-dark-secret-at-the-heart-of-ai/

http://www.invaluable.com/auction-lot/complete-r2-d2-unit-assembled-from-original-com-DBC4C5B95B

I had no idea that this is going on. Thanks a lot for posting that article. Amazing goings on.

Leave it to DARPA to be deeply involved. If anybody is going to unleash an out of control artificial being on us, it's going to be them.

LOL at this "“I think by all means if we’re going to use these things and rely on them, then let’s get as firm a grip on how and why they’re giving us the answers as possible"

duh.

There have been several instances of nuclear launches initiated by computers, that were prevented by a human being.

6 hours ago, Rich said:

This is a bit of a long read, but an interesting article on how the most successful / powerful artificial intelligence programs out there are self taught and due to the vast amounts of data that they compile and consume to come to their conclusions, no one really understands how they make the decisions they do.   Yet they are very accurate.

https://www.technologyreview.com/s/604087/the-dark-secret-at-the-heart-of-ai/

I would imagine they can calculate finite but wide range probabilities at an astronomical speed rate which dictates most their responses.

I think of it like a combo lock,  for a person it would take them a long time to go through the number sequences until they came up with the right number, for a computer, or AI it might be done instantly. The way we make a thought is instantly, but for a computer or AI, it might need to go through that combo lock process to determine which of those combo numbers carries the correct choice (or in most cases % outcomes). The highest probabilty is the choice it makes

22 hours ago, Rich said:

This is a bit of a long read, but an interesting article on how the most successful / powerful artificial intelligence programs out there are self taught and due to the vast amounts of data that they compile and consume to come to their conclusions, no one really understands how they make the decisions they do.   Yet they are very accurate.

https://www.technologyreview.com/s/604087/the-dark-secret-at-the-heart-of-ai/

Sounds like intelligence beyond our understanding. Logic in its purest form, in other words. That has the potential to be amazing and scary all at once.

30 minutes ago, blue_gold_84 said:

Sounds like intelligence beyond our understanding. Logic in its purest form, in other words. That has the potential to be amazing and scary all at once.

I think that will always be the difference between our logic and AI logic.  Unless you can program emotion and/or instinct into AI, all of it's logic and decisions will be cold and calculated. Human response varies on so many intangibles, people tend to do what's wrong even though they know better, or do what they do for other reasons they can convince themselves is for the greater good

32 minutes ago, Taynted_Fayth said:

I think that will always be the difference between our logic and AI logic.  Unless you can program emotion and/or instinct into AI, all of it's logic and decisions will be cold and calculated. Human response varies on so many intangibles, people tend to do what's wrong even though they know better, or do what they do for other reasons they can convince themselves is for the greater good

Agreed. Well said. Emotion and instinct make up a significant part of our thought processes, which I think is inherent in biological intelligence. Logic and calculation are no doubt present, but we're typically capable of explaining our rationale or motivation relative to what was thought or felt in that moment. For an entity whose intelligence is wholly based on calculation and logic, it's a different process and while the application to real world issues could prove useful, determining the why and the how is an obvious challenge and muddies the waters as AI advances are made. It's an extremely fascinating discussion, especially as technology continues to advance and play an ever growing role in our lives.

I found this last line of the article poignant:

Quote

...we should be as cautious of AI explanations as we are of each other’s - no matter how clever a machine seems. “If it can’t do better than us at explaining what it’s doing,” he says, “then don’t trust it.”

It begs the question: what if an explanation is beyond our understanding? I can't help but think AI would, if given the opportunity to do so, advance beyond our human intelligence. It seems inevitable.

I've seen the movie "Her" and I don't like where this is going!

42 minutes ago, blue_gold_84 said:

Agreed. Well said. Emotion and instinct make up a significant part of our thought processes, which I think is inherent in biological intelligence. Logic and calculation are no doubt present, but we're typically capable of explaining our rationale or motivation relative to what was thought or felt in that moment. For an entity whose intelligence is wholly based on calculation and logic, it's a different process and while the application to real world issues could prove useful, determining the why and the how is an obvious challenge and muddies the waters as AI advances are made. It's an extremely fascinating discussion, especially as technology continues to advance and play an ever growing role in our lives.

I found this last line of the article poignant:

It begs the question: what if an explanation is beyond our understanding? I can't help but think AI would, if given the opportunity to do so, advance beyond our human intelligence. It seems inevitable.

There's no disputing a machine can do calculations at a much faster speed than the human brain (with exceptions of beautiful minds maybe) but like the mind, there has to be a finite amount of knowledge it can intake based on storage space, so at some point I could see a conflict with the program.

I've heard (can't remember where tho) at a certain point,  the human brain needs to delete old memories to create new space for new knowledge/memories. Would AI follow the same suit once it hit it's limit, or would it simply stop growing? Or could it utilize USB stick type storage transfer? 

I got to thinking about the USB thing, and we kind of do a similar thing to that, but in our own way with pictures, videos, items that remind us of old memories. Ones we don't need to remember, but can go back to with the right trigger for that "oooh yeah" moment that may have been long forgotten.

All in all depending on the use of the AI, I don't think it's a bad thing, but maybe I watch too many movies, but I could see movies like the Terminator or The Matrix as a potential scenario if the program somehow ends up in survival mode. 

 

 

12 minutes ago, Taynted_Fayth said:

I've heard (can't remember where tho) at a certain point,  the human brain needs to delete old memories to create new space for new knowledge/memories.

The article says that the people creating the programs, don't know how they work.

If you don't know how it works, you can't assume anything about it, including whether or not it has emotions.

Sci fi preceding science.

This is Hal.

Edited June 13, 20178 yr by Mark F

4 hours ago, Atomic said:

I've seen the movie "Her" and I don't like where this is going!

http://www.mirror.co.uk/news/weird-news/man-builds-scarlet-johansson-robot-7667715