Serious discussion on AI

[large_ghostman]

Right so we all saw the link, and know what its about. Lets have a serious discussion on AI, not wiki posting or stuff like that but I want to know where you think AI will be in 5 years time.

I am asking because I had an idea of where it would be in 5 years (by idea I mean opinion), I knew about the young sex bot they used to catch pedos and a couple of other chat bots used in some countries to give advice etc. What I didnt know and hadnt seen was the research lab in Edinburgh, I obviously cant talk about everything I saw as I didnt ask what was and wasnt allowed. I know the robot that plays chess is ok because its been on tour, I also know of a couple of other things they have that have been made public, But I am aware of what i thought might be done in 5 years time is already surpassed in some of the stuff they have working right now.

Also is AI the right word? its been around a long time and its kind of stuck, but somehow when you see some the stuff thats being done I am not sure AI is the right word.

One other thing has anyone else seen the comments made at black hat regarding bitcoin software (the actual algorithm?), I am looking for the links again but more than one respected programmer who has seen the code has said its not of this world (obviously it is) but makes you wonder if the story of how it came about is right.

So no politics or moronics. AND NO ANT BASHING!!

 

[cowboybob]

AI, to me, is an attempt to replicate the functions of the Human mind, which will never be quantified. Nor, for that matter, qualified.

As such, its definition will never succumb to a definitive set of rules. Nor will an infinite set of rules be able to contain the concept such that a machine could be designed that could emulate Human intelligence.

I believe this because chaos ("... behavior so unpredictable as to appear random, owing to great sensitivity to small changes in conditions... ) seems to be the norm rather than the exception.

Which brings me to infinity. My concept of the term is a circle. I ask this simple question of people: "Where, exactly, on this circle is the beginning?". I would expect that at least one person would answer: "Who cares?"

I don't think any non-human intelligence, having been asked the question, would ever come up with that on its own...

 

[nsaspook]

Bitcoin first: Nothing special about it. I run a test core node at the house from the latest testing releases compiled from the source. The blockchain is very clever but the current implementation is showing some critical problems that need to be addressed.

AI: The little Tay AI bot experiment gone wrong shows how little we understand human intelligence when we try to mimic human intelligence with the equivalent of tinker toys when compared to the human brain.
https://www.washingtonpost.com/news...un-millennial-ai-bot-into-a-genocidal-maniac/
https://smerity.com/articles/2016/tayandyou.html

Intelligence is more than facts, computations, inputs and outputs. A human level of intelligence is interwoven into our very existence, life with other humans, feelings and an infinite number of other things we don't have a clue on how they work in a way that can be abstracted into machine circuits today. The fact we exist makes it possible for an 'artificial' version to be created but don't let the slick staged demos fool you into thinking we are any closer to the 'next 20+ years' AI time-frame.
https://intelligence.org/files/PredictingAI.pdf

 

[ronsimpson]

There are several computers I talk to frequency. One is for train reservations. That thing knows all the town serviced by am-track. It knows where each train is and when it will hit the next town. It understands my voice very well. It is just amassing how far computer have gotten.

 

[large_ghostman]

I would disagree about the block chain but dont have the reference to hand so will leave that for now.

I think part of the problem with AI like Tay is they are trying to get it to learn from human interaction, thats kind of flawed from the start. There will be enough people who think it funny and a bit of joke so they skew by knowingly take it down the wrong path, like the teaching the parrot to swear. The parrot thing is from experience, we had loads of birds in devon and had a parrot in the house. One of my dads mates when no one was in the room would teach it to swear, it got to the point the bird had to live in Avery its language was that bad!

Ultimately AI at the moment is still pretty linear, there is fuzzy logic involved but it still boils down to a choice system for response. The most complex system i have seen I dont know how it works and they wernt giving much away. This was the chess robot, interesting that it had 3 pairs of 6 cams mounted on it, one pair watched your face non stop. I was told to play a game of chess with it anyway I liked, so for the first game I thought i would try and catch it out, I played real noob style and made so many errors the game was over in minuets.

What was interesting was the second game, behind you where you cant see it is a screen, on that screen it shows the chess board, the computer flashes the square it thinks you will move too before you move. when the whole thing is played back you get to see the screen that was behind you, the second game i tried to beat the robot I played totally different to the first game and yet 98-99% of the time it guessed my move before i made it. I am pretty sure it reads your face and probably watches your eyes, but still really clever.

I asked alot of questions about it but they were not giving anything away. the other thing they have is a computer system that watches news feeds and the markets, then places 'bets' on stocks (not real money), its getting pretty accurate predicting market trends, if they used real money and actually played the market they would be making some serious cash. I think the market one disturbed me the most and i am not sure why.

They have all kinds of stuff there thats way way ahead of what i thought could be done, the voice thing you mention with trains is getting extremely good, I wonder what kind of computing power is needed to handle X number of calls at the same time and have no lag.

 

[Pommie]

I don't believe there will be artificial intelligence in my life time, Or the lifetime of my Grandchildren. Yes, there are systems that mimic intelligence but they are just very knowledgeable databases. As a programmer, I cannot see a path to intelligence. We still have no idea how the brain works. How could we ever surpass it?

Mike.
PS, is Serrie (sp? - apple girl) intelligent?

 

[nsaspook]

The predictions of AI happening 'real soon' are mainly tied to the human desire not to die. “The Singularity” seems to be directly tied into crackpot predictions of "Rapture"(fundamentalist Christian apocalypse).

Most of us believe in the Maes-Moore’s-Murphy’s Law: Immortality will arrive the year AFTER I die.

 

[large_ghostman]

I don't believe there will be artificial intelligence in my life time, Or the lifetime of my Grandchildren. Yes, there are systems that mimic intelligence but they are just very knowledgeable databases. As a programmer, I cannot see a path to intelligence. We still have no idea how the brain works. How could we ever surpass it?

Mike.
PS, is Serrie (sp? - apple girl) intelligent?

Without getting into the debate on the subject..... AI is alot like Global warming, from the very start it has been given a snappy but incorrect name. In the case of what is termed Global warming it might have proved better to call it climate change. Same kind of thing with AI, as you point out its not really intelligence as such but clever or rather sophisticated programming, but a truly intelligent system would surely require some biological component.

I can see AI getting to the point it appears intelligent but as with the chat bots if you say something it dosnt grasp your always going to get a stock answer to the effect of 'I dont have a clue what you just said'.
I think AI as we know and understand it is in a comfortable zone, i really wouldnt be a fan of a truly intelligent system, As I said before if a machine or computer was completely logical AND intelligent I dont think it would tolerate people for long.

I saw some really clever things at the uni the other day, but when you compare it with say ANTS (no ant bashing) then even top end AI cant compete with that they are able to do and achieve with no intelligence.
The one aspect of AI that does intrigue me is the swarm bots I saw, they are kind of modeled on ant like behavior, very very clever programming and really interesting to watch.

Then we have the whole tech growth is ex potential, so if you consider how long it took to go from caves to the industrial revolution it was a long time, but from steam to putting a man on the moon was very quick (in comparison), then from the moon to rovers on Mars is another giant leap. Even computers being rare 30-40 years ago? and now the average pc outcomputes a 1980's super computer. So I am kind of wary with time frames as tech has a way of jumping up a scale every couple of years.

I dont remember them but our tech teacher was talking about 9" floppy disks being a big break through then smaller ones holding 1meg of data being a big leap. I know we still have have a very old DX66 486 computer that was dads, it was considered a power users computer because it had a 1 meg video card and 8 megs of ram with a 120 meg HDD! Not sure of the age but I think the early 90's.
So maybe taking that as a bench mark another 100 years could be interesting times.

I still think it was asking for trouble getting it to behave like a teenage girl, I cant see the rationale behind that. Also giving it txt like speak and making it 'street' was always going to attract the wrong people. Shame they didnt try a couple of variations.

 

[large_ghostman]

The predictions of AI happening 'real soon' are mainly tied to the human desire not to die. “The Singularity” seems to be directly tied into crackpot predictions of "Rapture"(fundamentalist Christian apocalypse).

Most of use believe in the Maes-Moore’s-Murphy’s Law: Immortality will arrive the year AFTER I die.

Except as I have said before my grandad saw the birth of space comics that were seen as complete fiction and lived to see alot of it like flying to the moon become reality. And tech development gets faster and faster, although they will hit limits soon with wafer density etc, so I guess there will be a lag until that is sorted out. Nano technology is gaining ground but still not where it should be.

 

[nsaspook]

Except as I have said before my grandad saw the birth of space comics that were seen as complete fiction and lived to see alot of it like flying to the moon become reality. And tech development gets faster and faster, although they will hit limits soon with wafer density etc, so I guess there will be a lag until that is sorted out. Nano technology is gaining ground but still not where it should be.

That mainly assumes that we can 'grind' our way to the insights needed for some future AI by some exponential process of increasing computer power. Space comics are also filled with things like FTL engines and matter transporters that are really no closer to reality today or 100 years from now with any known technology. Things like flying to the moon were mainly engineering problems not problems with the basic science of mass, gravity and rocket equations that were solved long before. There is deep research into the basic science of intelligence that needs to be done long before Nano technology or any technology can be used to build the basic constructs from that science. As a old guy said once, "There are many unknowable unknown unknowns" before us.

https://cbmm.mit.edu/views-reviews/article/science-intelligence

In summary, we argue that the current focus on the engineering of intelligence, at its heart a problem in the design of better learning algorithms, must be complemented by scientific investigations of the only extant form of natural intelligence – that of homo sapiens. It is only once we better understand the biology of our own intelligence that we will be in a much better situation to understand the promises and pitfalls of artificial intelligence. In short, theengineering of intelligence, which is now attracting great amounts of resources and talent, calls for an equally grand and synergetic effort — on the science of human intelligence.