Bill Gates is worried about artificial intelligence too

AI exists.

The level that is alluded to here is the equivalent not of air travel, but of Jetson-esque flying cars that fold up into a suitcase capable of being carried by a skinny human. Space travel, not as we know it, but at light speed. When one of the first mainstream tablet computers was introduced (by Bill Gates), it was a flop. It took a decade for it to become a mass market viable product.

There will also be a data limitation for any computer system. There is a finite amount of space that can contain data. A robot observing its surroundings, and reacting, keeping track of its movements, and everything else will rapidly deplete any available data storage.

How much of what you did today will you recall in two weeks? For AI, the value is exact. Because otherwise, assume an upgrade to the system. What prevents that upgrade from being erased? Humans don't choose what to forget--we don't erase the ability to undergo cellular division. AI would have to actively choose what data to erase. In essence, they would have to actively define who and what they are, at all times.

The best an android could do would be to create virtual memory for potential upgrades, compile and run. If everything works absolutely, incorporate the code and move on. But a fundamental problem still exists--there is no way to determine whether or not a system is in an infinite loop. If a solution to that problem ever becomes possible, then I'd be concerned.

 

First two sentences here are definitely true, float. Not sure your last one automatically follows though. The limit could be pretty large. And couldn't AI machnes start constructing other AI machines, say, the size of the empire state building, with exponentially greater data storage capacity?

 

Or slaving other systems in the network.

 

Nah. Wirelessly connecting to cloud storage. It'll require fast wifi.

 

The FCC to the rescue

 

Think of how exponentially technology has grown and developed over the last fifty years, the last twenty, and the last ten. The potential development over the next twenty is mind boggling, the next fifty is beyond our full comprehension. If not, the only cause would be some sort of cataclysmic technology backlash that could occur in response to this very danger.

 

No it isn't. We refer to it as Moore's Law. We know exactly where tech will be in 15 years, and are already establishing ways to improve and get beyond that limitation.

 

Are you insinuating a hive mind, where one master computer does all the thinking for all the others? Or is there utility in having a computer the size of the Empire State Building? And even then, the amount of data is still finite.

You cannot build an infinitely large system, which means there is always a limit to data, which means that AI will always have to determine which part of memory to actively delete, and that means they are always in a memory deletion loop, and therefore the chance of error is always present.

 

Hell, in order to preserve an entity the size of the Empire State Building, there would have to be a minimum equivalent sized or proportioned memory block just. for. the. backup.

And anything that is capable of making a backup of itself, and does not, is doomed to extinction.

 

I feel like you are not using any sort of imagination, here. An AI could create sophisticated ways to compress data and increase efficiency. It could ration memory while manipulating people into building more memory or eventually automating that entirely.

There are already worms and viruses that combine computing power from many separate systems through a network, sending information to and from a central location. Moore's Law won't be very applicable as quantum computing or even alternatives to silicon semiconductors comes along. Moore's Law is simply an observation considering the spatial limitations of continuing to shrink down the space needed for x number of transistors and such. As technology moves away from that, it won't be as relevant.

http://www.wired.com/2015/01/the-rise-of-diamond-technology/

 

Why are you assuming the half of the building couldn't be the backup? You're awfully hell-bent on acting like Stephen Hawking and Bill Gates are fools.

 

JQK is going the way of the Dodo bird.

 

I would be most interested to see exactly where we will be. Can you enlighten me? I have never seen any of these projections of the future be all that accurate. You ate basing this opinion on known data and capability. Major breakthroughs are seldom known until they happen. It is this very attitude that will allow such change to occur.

 

And the comparison to flight and moon landings is really not a good one. We understood flight and gravity.

We have no freaking clue what consciousness is nor how to even begin making it. Which is why I think the first conscious AI might be a complete surprise if it even happens.

And as for memory deletion, that will be a requirement as float says. He'll, it might be a requirement for consciousness or the system might break down from lack of "sleep".

 

Moore's Law will always be applicable because the smallest you can ever go is atomic. Ergo, there is always a limitation.

Changing materials does not alter that fundamental physical property that the smallest anything can be is finite.

 

I'm not, I'm hell bent on people taking conjecture as fact. Time travel is something heavily debatable, mentally believable, arguable, and physically unlikely to ever prove possible.

This SCIFI AI dominated world is the same crap. There are absolute, unsurmountable limitations.

 

As far as we are capable of thinking at this time.

 

I already have. Ferroelectric will be the next step for general consumption. It is cheaper than diamond and out paces Silicon. It has been known for a while. It has been modeled and shown to be way more efficient quite recently.

It has known as a "major breakthrough" for a long time, and I imagine that someone will wake up in a few years and think it just occurred recently. It didn't.

 

No, zero will always be zero. Infinity will always be infinity.

You can't go smaller than zero, therefore there is always a size issue. You can't read infinity, therefore there will always been a read issue. And since you can't know where you are on a circle, you can't detect if an infinite loop is a loop or a process.

 

Just as we cannot know the exact future. It is smug assurance which will lead us to our doom.