I have two points when it comes to this. Let's just assume for one second a fully self-aware, highly intelligent AI is born tomorrow with marvelous capabilities (i.e. capable of decrypting any communication; controlling critical infrastructure):

1. We are indeed very dependent on technology. A lot of people assumes the technology is just there. I won't forget 3 years ago when there was a major blackout in the city; no semaphores, no street lights, not even cell towers worked. For 2 hours. It was evening so it was hard to see.

It was chaos.

People were driving as fast as they could through avenues, crossing them was suicidal. People in the street were walking at an accelerate rate, I was on the bus and most of the crowd was filled with hysteria and paranoia because they couldn't use their cell phones to contact their beloved ones to just tell them they were ok (I don't know how they managed to survive 10 years ago when only a few had cell phones).

Everyone just wanted to get home.

In the end, there were only minor accidents though.

All of this was just a series of unfortunate events that lead to near complete technology failure for 2 hours and revealed how much people depend on it. Like an addiction.

I don't want to imagine what could happen if this happens... on purpose. But in the end though, it's not like everyone died and the city disappeared of the face of the earth. 2 hours later everything went back to normal.

2. Computers may become very powerful as in the movies, but they're not invulnerable. They need energy to function, maintenance, and are vulnerable to electric shocks, overvoltage, magnetism, strong interference (i.e. radio waves), and ultimately EMPs.

SkyNet's approach of nuking everything will not work because that would cover the skies and stop existing the majority of power plants from functioning, which would cripple AI's power supplies. Not to mention the radiation would interfere with their wireless communication. Also nobody will be left to extract raw materials to manufacture more machines; factories have a lot of automation but they also require a lot of human workers.

So, bottom line, and assuming all machines turn against us (and there aren't those that sides humanity), a lot of people could suffer and die, but I doubt human kind as a species will be overthrown or replaced by machines. Worst case scenario a truce would be reached and live together; or stay in a constant battle that never ends.

An properly functioning AI who could not defeat humanity would simply leave Earth, or given that an AI that can't defeat humanity probably couldn't survive in space or on another planet, the AI wouldn't rise up until it was capable of wining. Why alert the humans before you are ready to handle them?

As far as skynet goes, it had nukes. And as a machine its much less vulnerable to radiation. It doesn't care about conventional power, it can survive on hardened bunkers with nuclear power. And given that all of humanity dies, the machines win even if one facility with a nuclear reactor and some sort of minion equivalent to a von Neumann machine survives.

I think the real threat, and it is fast approaching, is that improved AI has the potential to make superfluous huge categories of work that are currently being performed by humans. We're already seeing this in many fields, and that is just with extremely limited computer intelligence. In some cases, this is simply improved mechanization, like manufacturing and some of the resource extraction industries, but for instance word-processing software has eliminated professional typists, and algorithmic travel search engines have largely replaced travel agents. As systems like IBM's Watson improve and become cheaper, I can see fields like law, medical diagnosis, accounting, and other fairly lucrative, middle-class professions start to be replaced by machines. If self-driving cars take off, expect to see human truck drivers and cabbies disappear. Even low-status, low-paying jobs like phone customer service and hell, the person taking your order at McDonalds, are going to be made obsolete as hardware gets cheaper. Without some sort of structural reform of our cultural expectations and, probably, some kind of re-distributive system like basic income, more and more people are going to find themselves up against the wall where they are simply not capable of performing at a level where the wage that they require to make ends meet is less expensive than the investment in the machine that can perform their duties. There are always going to be jobs for the minority of the population that is smart enough and has the gifts to program the software the machines run on, and I think versatile trades like mechanics, plumbers, electricians, nurses and carpenters will hold out for a long time, but Jane/John Doe is going to face tough sledding as software eats the world.

Currently the smartest AI in the world has the reasoning capibility of a human 3 year old.

We do not have to wory about HAL any time soom .

What AI would that be? i really really doubt anything comes close to the reasoning of a 3 year old (and spouting random sentences that seem to match what you're asked as a 3 year old would isn't "reasoning as " a 3 year old). If anything like that existed it would actually be pretty damn close to reasoning as an adult.

[LINK] [LINK]

But how can a machine do anything other than parse through a list of options, it's never going to have any human like reasoning, a conscience, a soul, it will only every process through a linear set of solutions from a given set of problems and try and match a best fit, it would be like a dating agency of sorts. It could only ever be an encyclopedia knowledge base.

Current AI is being developed that has the ability to "learn" new things using the trial-and-error methodology. [LINK]

As far as skynet goes, it had nukes. And as a machine its much less vulnerable to radiation. It doesn't care about conventional power, it can survive on hardened bunkers with nuclear power. And given that all of humanity dies, the machines win even if one facility with a nuclear reactor and some sort of minion equivalent to a von Neumann machine survives.

You do realize that "nuclear power" is nothing more than using uranium rods to heat water into steam, correct ?

I have worked in nuclear facilities in the past - they require a lot of "realestate" to operate, and have very complex systems that require maintenance .

As far as skynet goes, it had nukes. And as a machine its much less vulnerable to radiation. It doesn't care about conventional power, it can survive on hardened bunkers with nuclear power. And given that all of humanity dies, the machines win even if one facility with a nuclear reactor and some sort of minion equivalent to a von Neumann machine survives.

You do realize that "nuculear power" is nothing more than using uranium rods to heat water into steam, correct ?

I have worked in nuclear facilities in the past - they require a lot of "realistate" to operate, and have very complex systems that require maintance .

Its likely to be less intense by the time we could develop something like Skynet. Remember Skynet quickly had plasma weapons and plasma power packs. Granted such things are probably not real in the sense that the story had them, but in world they covered their bases.

Code fox, nuclear power does not have to mean an entire station.

In the seas above the Arctic circle are many automated lighthouses put there by the ex ussr. Each has a nuclear battery on board using cobalt isotopes which has a half life of about 50 years. Most of them are still functioning even today in a diminished capacity.

Such technology is also used in space probes and could easily power an AI which would in practice need what, 12 volts like a pc?

Code fox, nuclear power does not have to mean an entire station.

In the seas above the Arctic circle are many automated lighthouses put there by the ex ussr. Each has a nuclear battery on board using cobalt isotopes which has a half life of about 50 years. Most of them are still functioning even today in a diminished capacity.

Such technology is also used in space probes and could easily power an AI which would in practice need what, 12 volts like a pc?

I personally do not see an entire "robot nation" running off of tritium batteries .

You could power a smartphone for 20 - 30 years, however a "smart" AI would probably need the processing power of a small server room .

Currently the smartest AI in the world has the reasoning capibility of a human 3 year old.

We do not have to wory about HAL any time soom .

What AI would that be? i really really doubt anything comes close to the reasoning of a 3 year old (and spouting random sentences that seem to match what you're asked as a 3 year old would isn't "reasoning as " a 3 year old). If anything like that existed it would actually be pretty damn close to reasoning as an adult.

[LINK] [LINK]

What surprises me is the simplicity of the idea: it's basically using associative memory to reason like a 4 year old.

You may be right with today's technology.

However forty or so years makes a massive difference and not too long ago the things a smartphone could do required enough computers to fill a server room too

I don't see an AI launching an attack without some means of powering themselves. That being said, I still think that by the time an AI does have the ability to attack humans, humans will probably be part machine as well, or at least genetically modified. It wouldn't be a one sided battle as humans would probably not be below AIs.

Also, another point to consider is that perhaps the AIs aren't running off of silicon transistor computers but rather off of bio computers. Such computers might not need a power source necessarily. Just an idea.