If the group has a negative social mood, believing that tomorrow will be worse than today, the bias goes in the opposite direction. Instead of "welcoming" we have "rejecting," instead of "global" we tend to see events that are "local" and so forth.

From the 1990s onward, the financial sector created a vast array of instruments designed to separate investors from their money, financial derivatives of an ever-increasing level of complexity. At some point, this complexity reached a point where even the creators of the derivatives themselves didn't understand them.

The global financial system consists of firms in the financial services sector - banks, hedge funds, insurance companies and the like - and various governmental agencies who are charged with regulating these firms.

System theorists know that it's easy to couple simple-to-understand systems into a "super system" that's capable of displaying behavioral modes that cannot be seen in any of its constituent parts. This is the process called "emergence."

If you ask which of the scenarios I think is most dangerous, though, I will give a different answer. In that form of the question, I regard a nuclear attack, terrorist-generated or otherwise, as the most threatening combination of likelihood and long-term damage to modern life today.

My own guess is that quite quickly the machine intelligence will start dreaming machine dreams and thinking machine thoughts, both of which would totally incomprehensible to us. This would then lead to each species, we and the machines, moving off on to its own separate life trajectory.

I think that in the immediate aftermath of a superhuman machine intelligence revealing itself, most people would feel very threatened but take solace in the thought that we can always pull the plug.