Terminator

[007]

Just watched the original Terminator (again). I really like these movies. Lots of explosions & car chases. The robots rise up
and take over in the year 2029 (approximately).

My question is : Does anybody really think this could happen? I have been mulling it over & I am not sure that it so far fetched. I mean, I'm not sitting here worried about the robots, but is it possible that eventually the machines will rise up & kill off the human race? Could machines achieve this state of artificial intelligence? Would they want to kill all humans? Could this happen by the year 2029?

Has anybody seen I Robot? It was pretty good. I like the premise of that movie & it seems to have a bearing on this subject.
What do you people out there think? Could this happen?

[slam]

I think any AI that could conceivably challenge the combined efforts of the human race is a considerable ways into the future. We may achieve some limited AI goals, such as autonomous robots that can walk and move with perhaps super-human abilities, or systems that are even better at recognizing faces than we are, but trying to mimic even once brain is a monumental task. We've still barely scratched the surface in learning how the brain works, let alone how to program one.

One of the trickiest things to overcome will be the fundamental differences in how our brains and our current computers work. Computers just step through instructions one (or maybe 100) at a time. Sure, they can go really, really fast, but it's still a finite amount of work at any given time.

The brain fires off it's little neurons and relases chemicals which can affect thousands or millions of other neurons at a time, creating an incredibly complicated web of billions of interactions all happening simultaneously. it's incredibly more effiecient, but also near impossible to decipher how it works.

There's the possiblity that we could maybe come up with superior AI without having to understand our brain, though. Evolutionary AI just goes as it the same way nature does. For example, create two programs that play backgammon. Play them against each other, and delete the losing program. Copy the winning program, introducing some random variations, and play the two new programs against each other. Rinse and repeat a few million times and you can end up with the best damn backgammon player in the world, without ever having to even learn the rules of the game.

I'll know we're near the end when robots can design themselves. That's when we need to blot out the sun with nuclear winter (wtf? matrix sux)

[matty]

These robots you fear are currently under construction in my secret Evil Robot Factory. Your doom is near, bitches....

[larryhead]

It's already begun!

http://news.bbc.co.uk/1/hi/sci/tech/4714135.stm :escape:

[007]

Yes, LH, but I ask you : Does Q1 know how to make homemade dynamite or fire an uzi? Also, can she repair herself using only an Exacto knife?

[larryhead]

2029 is only 24 years away, but in terms of technological change, that is an *enormous* period of time. It's really too far away to predict with any accuracy, but things will look a lot different than they do today. Will there be AI? Definately. Will it be like Terminator? Probably not.

Think back 24 years to 1981. Most homes didn't have a computer and I don't think cell phones were even around then, hell you had to call people from home... and there were no cordless phones. No caller ID. For gaming you had Atari 2600. No one used CD's or DVD's or MP3's. You bought tapes. No digital cameras or camcorders. The internet was just a few connected university nodes, used for research purposes. I don't think anyone could have predicted what 2005 would be like. Now all that infrastructure is in place and I think things will be changing much faster during the next 24 years, than the previous 24.

Moore's Law
Gordon Moore, head of Intel in the mid 1960s observed that the size of each transistor on an integrated circuit chip is reduced by 50 percent every 24 months. The result is the exponential growth of computing power over time. Moore's law doubles the number of components on a chip as well as the speed of each component. Both these aspects double the power of computing, for an effective quadrupling of computer power every two years.

Ray Kurzweil extrapolated back in time, and discovered a generalisation of this law holds even as far back as 1900. He extrapolated it forward and discovered we can expect a computer with the processing power of a human brain around the year 2030 in a computer costing $1,000.00


My guess is the AI then will be similar to now, being oriented to a specific task (ie. like the AI in your car or the AI inside of "smart homes"). :expert:

[Diamond Vision]

I think it's important to always keep robots programmed with the sense of being your bitch, no matter how strong, capable or superior they may be, they will always be our bitches. Then we can always treat them like shit and never worry about them rising up to destroy us; because they think they're our bitches and will never know any better.


stoopid robots.



along the same line, in the Oz books, there's a character called Tic-toc, the mechanical man. He was requsitioned by the King of Ev because he liked to beat and whip his family but instead he would whip tic-toc and he would never feel pain. The whip simply kept his copper hide shiny. He liked it. what a submissive fucker, eh?