Tuesday, August 11, 2015

The machine apocalypse, oddly.

According to computer scientist Donald Knuth, "artificial intelligence has by now succeeded in doing essentially everything that requires 'thinking' but has failed to do most of what people and animals do 'without thinking'; that somehow is much harder." It has come to the point machines are reaching new heights in reasoning, but can not temper their intelligence in any natural manner. It is possible to program a computer to play chess, but impossible to teach it the empathy to let a small child win once in a while to keep them interested.

When I need to go somewhere, and need directions I just ask my phone. "Hey, Siri, how do I get to the North Market on foot?" Siri tells me. Step, by step. Siri knows how to get me almost anywhere. But, has no clue that I can get out of the building. Most people might say, "go to the stop sign, and head west."

Computers, robots, and their controls are insinuating themselves into every aspect of our lives. We read on screens that get the text from cloud services, transmitted via towers controlled by a network of computers.

Traffic patterns are regulated by algorithms designed to keep things flowing smoothly, and cameras record the results, which are stored on and analyzed by a machine. It would be almost impossible to list the many ways and places technology shapes and controls our lives. It would be equally futile to claim we don't benefit immeasurably from the help.

It becomes increasingly obvious that we are more reliant on the convenience of our electronic servants daily. They assume more control, and more responsibility for our every day successes and accomplishment. But, things are not always fool proof, either. Anybody who has spent hours working on a document, presentation, or spreadsheet only to lose it, whether from computer crash, misplaced file, or programming "bug" can tell you about the frustration involved in digital manifestations.

It is impossible for even the best programmers to account for every possibility.  One doomsday scenario from Nick Bostrom in "Superintelligence, Paths, Dangers, Stategies." imagine a computer instructed to maximize the production of paper clips. As the machine interprets the instructions it may not be able to make the assumptions that come from years of evolutionary reliance on restraint. In it's quest to complete the program it would eventually convert all of the Earth's resources into office supplies.

Yes, that is an extreme example, but not impossible. And it is only one of many potential catastrophes. And if we know anything about mathematics it is that anything that carries a statistical probability, no matter how minute, that doesn't diminish over time will eventually happen. When the rapid escalation of technological innovation, and our reliance on technology is added, the probability increases.

In future posts we will cover all sorts of potential calamities. There are millions, and it will make a great movie, don't you think?

2 comments:

  1. In lies the problem that most good science fiction anticipates. A unit that is built and can only process logical algorithms can only find it's way in best practice to the logical end point. If we create something that is able to "think" then the logical conclusion that it would have to come to is that humans are a factor in almost all problems.That can't end well.

    ReplyDelete
    Replies
    1. Which proves that machines can be pretty darned smart,

      Delete