What is the Kobayashi Maru problem?

The Kobayashi Maru is a classic example of a hard problem in artificial intelligence: Suppose you are the commanding officer of an Enterprise-D starship, and your job is to guide a new crew, who have just been through training, into a situation in which they will be forced to choose between right and wrong.

In order to be able to decide which option is the right one, you need to be able to make a prediction about how your crew will behave. You could spend a lot of time with your crew to get to know them and their personalities, and it might be possible to predict how they would react. Or you might take a shortcut and simply send them a mission that will show them what kind of action would be wrong, and that they can expect to be punished for it.

Now suppose that you send a group of trainees on the mission, but that they all reach the same conclusion and report that it was wrong. How do you know that the mission was not just too easy for them, or that the training was not deficient? The solution is to give them a mission that is very difficult, and for which there is no correct answer. The hardest test you can think of is the Kobayashi Maru, an impossible mission that has only one right answer. No matter how well your crew trains, they will never be able to predict how they would have acted under those conditions, and so you cannot trust them to tell you what the right thing to do is.

Artificial Intelligence researchers have been interested in the Kobayashi Maru problem since it was first proposed by Stephen C. Kobayashi and Edward L. Wolff in 1958. (As you might expect from a name like this, the problem has also been known by other names: The Dilemma, The Impossible Mission, The Trainees' Dilemma, The Impossible Mission Problem, and The Training Problem.)

Since it was first proposed, the Kobayashi Maru problem has been a hot topic of discussion among Artificial Intelligence researchers. However, over the years, the problem has become more and more obscure, and today most people who hear the name know that it has something to do with computers. This article attempts to explain what the problem is, and what it has to do with Artificial Intelligence.

How did Scotty beat the Kobayashi Maru?

Let me count the ways.

A couple of months ago I was at a bar where a friend and I got into a discussion about Star Trek. After a few drinks, we were talking about a conversation I had with another friend at that bar who told me he was a big fan of Star Trek. This led to us comparing our knowledge of Star Trek with each other. It quickly became apparent that my friend didn't watch many episodes, while I had watched all of them. He asked what episode I was watching next. I replied TOS, 'The Enemy Within', then 'Space Seed'.

He asked if I knew that episode. I replied yes.

He told me that episode is one of his favorites. I said it was indeed a favorite.

He then asked if I knew the Kobayashi Maru. I told him I did not know that episode. He laughed and told me You're a liar. We had a great laugh. He then told me that he was going to watch 'The Kobayashi Maru' and give me a report on it. We then sat down and had another drink.

As were waiting for our beers, he told me that the episode was a great one. I told him that I didn't know that, but I'm sure it's good.

He then told me that the reason he liked the episode so much is that it shows how human Kirk and Spock really are. I had no idea what he meant by this. Then he told me that when they had the briefing in 'The Kobayashi Maru' they had to make a test to determine whether the ship was capable of getting itself out of the simulation. In other words, it was a test to see if the computer was able to tell the difference between reality and a simulation.

Wow, I said, That's clever. No, he replied, Not really. The ship had to pass the test.

I didn't understand this point. Then he told me that he was reminded of this in an incident from his life where he used a computer.

Has anyone passed the Kobayashi Maru?

I know most don't, but I am taking my first test on the Kobayashi Maru on Saturday.

I've done a lot of research and there isn't a question to be found online asking whether one has passed this test or not. Does anyone here know if anyone has?
I don't know much about test taking, but I don't think you need to know that they didn't change the final answer until after you have passed the test. You could just tell them "pass."

This whole thing is so funny to me. I remember when I was taking my test it was one of those questions they ask you like 'Do you think the person who says "Kibai-shi" can fly?' It makes you want to just laugh. There are a lot of jokes that I'm missing in Japanese right now. I can't pass it either time for whatever reason so I just smile at it.

Yikes - I didn't realize there's even such a thing as a 'wrong' answer! This post just changed my attitude about testing. I haven't taken any but my sister has taken hers twice, once as an undergrad with a 4.0 average (in case you're wondering she's a psychology major) and the other time as an MBA grad student. She told me once you get that weird green light that says "test over." I guess she got that one. I'm interested to see how I do when my test date finally arrives.

The last test I took there was one where you had to say "halo halo" instead of the correct answer "bam bam" (you probably guessed I was a sports fan).