Artificial Intelligence already creeping into our society (part 5)

It may seem that I am simply focusing on examples of computer errors in all this, and not actual AI. Yet Artificial Intelligence is a conglomeration of complex algorithms that allow a computer to make conclusions based on data sampled through various means. What we might call a computer error is really a matter of perspective—to us, it is an error because it violated our intent for the computer. But in the cases mentioned this week, the computer arrived at that “error” through a logical application of its programming. The AI on the International Space Station had a task to complete—the launching of satellites on a specific schedule—but when the AI realized it could not maintain its objective if it was delayed any further, it simply resumed its task, regardless of the fact that it had been told to stop. The peak-rewards situation was a mere application of programming to a situation: it was hot, everyone was using their AC, and the computers decided to shut down the AC of everyone on “peak rewards” because too much electricity was being used. I've even heard a story of a hospital situation where orders for medicine suddenly stopped being printed out, and pharmacists did not realize that the queue was building internally in the computer system. Patients no doubt suffered for the delay in their medication, but because of a computer program re-routing the notifications to a computer instead of a the usual printers, the pharmacists were delayed as they tried to figure out what was going on. Why are we even considering developing AI for any system that could profoundly affect our lives? Smart phones, smart homes, and Google's smart car...does AI make you feel safe?