Computers learning to ‘regret’ decisions, routing paths, and serving humanity

hal
I shouldn't have done that, Dave.
hal

I shouldn't have done that, Dave.

In Tel Aviv, researchers are working toward integrating computers with ‘regret’ in order to improve overall efficiency.

In theory, a computer system would register the difference between a desired outcome and the actual outcome, then attempt alternatives over time to achieve or exceed the desired outcome without instructions by a human user.  This alone is nothing new; the difference is in the integration of a value, regret, which the computer system is motivated to avoid or reduce (“how?” is the question I pose, as though scientists could psychologically bully a computer into being disappointed or ashamed – but I guess it’s just an analogy anyhow, right?)

Though perhaps the first step toward SkyNet becoming self-aware, the research is nonetheless funded by Google and promises to increase the efficiency of everything from bandwidth usage to processor speed.  [Source; Image]

Categories
HeadlineScience and Tech
No Comment

Leave a Reply

*

*

Editor's Picks