The apocalypse revisited – aggiornamento easy ways to make money on the internet!

You may have read recently about artificial intelligence — whether it’s going to take our jobs, be used in deadly autonomous battlefield weapons or exacerbate inequality easy ways to make money on the internet and drive the fruits of the economy further into the easy ways to make money on the internet hands of the rich. These are all reasonable things to worry about. But there’s another concern: that AI will become as clever as we are; that, when it does so, it will become as good as we are at designing easy ways to make money on the internet AI systems; and that it will recursively improve itself in a sudden easy ways to make money on the internet explosion of intelligence. And then, once we are trailing in its intellectual wake, it will exterminate us, not out of malice or hatred but because we are easy ways to make money on the internet inconvenient.

Something truly superintelligent, if we can build it, might be capable of doing things as far beyond our easy ways to make money on the internet capabilities as the cassini probe is beyond the twig. And if we’re not extremely careful about how we build it and easy ways to make money on the internet what we tell it to do then, according to this concern, we could face human extinction. Alternatively, if we get it right, we could be looking at a glorious future, guided and protected by a benevolent AI deity.

The rationalists are not the first to worry about this easy ways to make money on the internet sort of thing. Irving good, a british AI researcher and colleague of alan turing, foresaw the possibility of the “intelligence explosion” back in 1965. He said that “the first ultraintelligent machine is the last invention that man easy ways to make money on the internet need ever make. Provided,” he added, “that the machine is docile enough to tell us how easy ways to make money on the internet to keep it under control.”

Recently, more people have started worrying about what AI might do easy ways to make money on the internet when it becomes really clever. All the ais that exist at the moment are “narrow”. The chess app on your phone could probably beat a easy ways to make money on the internet grandmaster but it can’t help you cook dinner. Google maps can tell you, almost instantly, the quickest route to basingstoke but it does not know easy ways to make money on the internet the fischer defence from a king’s gambit.

He wrote up these fears in a series of blog easy ways to make money on the internet posts that became known as the sequences. The blog developed a following; the following became a community; the community, in time, became known as the rationalists. They developed real-life hubs, the largest in the bay area of san francisco and easy ways to make money on the internet in new york, but also in other cities around the world. Nick bostrom at oxford university’s future of humanity institute is working to make sure easy ways to make money on the internet that the first AI is ‘friendly’

The rationalists are an odd bunch — nerdy, intense, usually bright, deeply strange. And they have some unusual practices; lots of them, for instance, live in group houses and practice polyamory. They’re controversial: they are accused of being linked to alt-right weirdos and scientific racists (these links do exist but they’re not representative).

The rationalists’ fundamental concern is that almost any goal you give a easy ways to make money on the internet sufficiently powerful AI, however seemingly innocuous, could go terribly wrong. The fear is not the terminator: AI achieving self-awareness and deciding to attack its creators. It is the broom in the sorcerer’s apprentice: something doing exactly what you asked it to do but easy ways to make money on the internet taking you entirely, disastrously literally and you not being able to stop it.

We assume that intelligence will be like human intelligence but easy ways to make money on the internet there’s little reason for that assumption. A large part of the rationalist project is dedicated to easy ways to make money on the internet examining the ways that our brains, with their biases and kludges and shortcuts, are different from true, optimally rational, thinking machines — and while an AI might not be perfect either, it’s unlikely that it will be imperfect in the same easy ways to make money on the internet ways that we are.

For instance, imagine a computer that is told simply to become good easy ways to make money on the internet at chess. As the AI researcher steve omohundro points out, for pretty much any goal you give an AI, there are certain things that will probably be useful. Becoming more intelligent by building more memory banks and processors easy ways to make money on the internet is one such thing. There’s no theoretical limit on how much computing power you easy ways to make money on the internet could need: you can always be better at chess. So it will always need more material out of which easy ways to make money on the internet to build hardware.

You might think that you can switch it off, or reprogram it, but the AI may not want to be switched off easy ways to make money on the internet or reprogrammed; it won’t win many chess games with its plug pulled out easy ways to make money on the internet or with a new line of code telling it to easy ways to make money on the internet care about draughts instead. So if your AI is powerful enough, it may resist your attempts to switch it off quite easy ways to make money on the internet strenuously, perhaps with nuclear weapons.

A 20% chance is not the same as “it will happen”. But we don’t think it’s silly to worry about much smaller risks than that: there’s only about a 1 in 200 chance of my easy ways to make money on the internet children dying in a road accident of some kind but easy ways to make money on the internet I’m happy for the government to spend money on traffic-calming measures. Even if there’s only a 1% or 2% chance of complete disaster in my children’s lifetime, it still seems worth spending some time worrying about.

That’s what yudkowsky and the rationalists are doing at the easy ways to make money on the internet machine intelligence research institute in california; also nick bostrom at oxford university’s future of humanity institute; and max tegmark, a professor at MIT and one of the founders of easy ways to make money on the internet the future of life institute. Elon musk’s openai, and the open philanthropy project funded by the facebook billionaire easy ways to make money on the internet dustin moskovitz, are doing work on the same idea. That is, to make sure that the first AI is “friendly”, that it cares about what we care about and wants easy ways to make money on the internet to protect us.

Even if they achieve these aims, there are complex and seemingly intractable problems to be solved. Will humans enjoy being relegated to second-class status, little more than pets to the godlike ais we create? After all, we often think that we find meaning in our lives easy ways to make money on the internet through overcoming challenges; if we can simply ask the benevolent AI deity to easy ways to make money on the internet overcome the challenges for us, will that disappear? And will the ais themselves be conscious and therefore deserving easy ways to make money on the internet of moral consideration, or will they just be tools?

RELATED POSTS