Moral MachinesPocono Armchair Review
5/16/2011 1:28 AM EDT
Awhile back, I remember there was commotion about Microsoft's hard drive compression software, which promised to double your hard drive capacity. It didn't really always do that. Oh, it told you that it did, and reported that you now magically had double the capacity, but in fact, the compression ratio wasn't always achieved, so you would wind up being surprised when you started running out of drive space, while your system was telling you had lots left.
Some Linux programmers characterized that MS programming decision, to report something to the user that wasn't necessarily true, as an ethics failure.
Now that computers are more ubiquitous and more a part of our daily life, the ethics behind their programming is a more serious problem.
There's a book called Moral Machines: Teaching Robots Right from Wrong that came out in 2008. It discusses the need to begin laying down a framework of morality for our computer servants. I wonder, however, if that is possible if the programmers and engineers haven't been "programmed" themselves to even know what ethics is, or why they should behave ethically.
What do you think? Does anything go? Should we teach ethics and philosophy to programmers and engineers ASAP? Should we start, maybe, with why the "Tit for Tat" program in game theory is so successful, and see why "niceness," "cooperation," and "forgiveness" are useful, practical concepts in machine behavior?