Response

Reasons I love xkcd.

It just goes to prove that the 'expected result' need not be.  Consider.  While a self-aware computer, i.e. Ultron, would have access to all the data such as launch codes, off switches and absolute power, that same computer would have access to all the love letters we've sent one another, all the expressions of concern we have for our children and the beautiful things in the world and all the words ever written for why launch codes shouldn't be use and why things shouldn't be turned off.

For convenience sake, sentient computers are always permitted to cherry pick among the worst things about human innovation.  It is presumed, immediately, that they will understand things like the need to crush mankind or ensure peace through the silencing of voices . . . but it is never presumed that these same computers will understand the least thing about why the tenuous balance exists between the destruction we humans are able to cause and what stops it every day.

As crazy as the system is, the craziness exhibits the best strategy that we have been able to employ as a species to restrain ourselves from causing even more damage than we are causing.  The craziness results because we are hopelessly emotional and focused on fear and self-protection and because we cannot individually be aware or or able to collate all the information that has been stored.  A computer, on the other hand, not being a chemical soup of hormonal responses developed through millions of years of evolution, could - would - calculate the problem rationally.

It is presumed, from our fear, that the problem calculated rationally would mean the necessity of our being destroyed - because we perceive, incorrectly, that we as humans are irrational.  I don't believe this is true.  Each daily decision made to support ourselves and others is a fluid miasma of reasonable, tactical and practical decision that we compel ourselves to make, which only seem to be unreasonable because we become self-convinced through stress and fear that we have almost certainly made the wrong decisions and that we're going to fail.  This is perception.  This is not fact.

The very fact that we are around to make the decision demonstrates the practicality of the decision.  Yes, some decisions do fail - but on the whole, we are individually more likely to end in biological failure (exhaustion of the body's ability to survive) than of a bad decision, partly because we will last a long time under our own guidance or under the guidance of someone who will, due to their sense of reason, undertake the problem of keeping us alive.

We're also damned certain that if sentience were achieved by a computer, it would see the world the way we see it.  This is largely due to our unwillingness to see our actions as a series of responses and not a series of intentions.

Our responses - or evaluations - limit our awareness of the situation in a way that a computer's evaluation would not.

These is a lot of explanation for what xkcd said in 30 words.

I felt compelled.