As far as I know it is not possible for a computer program to write jokes.(or no such program exists as of today)
This sounds obvious and ridiculous at the face of it,
however my question is WHY is this not possible, what is the missing “piece”? Does it have to do with artificial intelligence?
It wouldnt somehow be possible to feed a computer 1000s of jokes, make it learn what is funny, construct a program around it, and make it write own material?
ps. the jokes doesn’t have to be “good” (subjective). Just something that most people would recognize as being humorous.
You’ve got to realize that most computers don’t understand language, and a grasp of language is necessary to construct humor, as you have to be able to understand how someone else will perceive what you are saying.
About the only other way to make computers do jokes would be to vastly limit it, so that it’s really just following a joke formula and replacing words. And, even then, without something like that Watson computer, it doesn’t really have a good way of knowing which jokes make sense.
In my view humans react to different situations with different emotions: happiness, sadness, fear, boredom, etc… Humor is our “none of the above” reaction. In other words, something is humorous when it lies outside the categories of things where other reactions are appropriate. That’s why it’s so difficult to reduce humor to bits and bytes. To grasp whether something is humorous, you have to understand not just one emotion, but all emotions. Furthermore, things become more funny as they get farther and farther away from other categories, which is why there’s a special ‘zing’ required for truly great humor. No one quite knows why it’s funny to watch two guys arguing over whether a parrot is dead, but it just is.
One first has to understand what makes something funny. IMHO the major item is having an expected pattern and then having the item not only be out of that pattern, but recognize an entirely different and unexpected pattern that also fits. A reorientation to the data set, a perceptual flip, at a higher nested processing level than at say switching figure ground in perceptual illusions, then occurs. If in the process it runs against subjects that members of the culture are uncomfortable with (such as sex in American humor) then that slight discomfort adds to the perception as “funny”. Also sometimes if it allows the listener to self identify as a member of an in-group (the inside joke) humor value is added. All these things require parallel processing pattern recognition, and understanding cultural norms, a still difficult tasks for AI methinks.
To that way of thinking Monty Python is funny because the pattern should not be selling a dead parrot. It takes a culture as traditional rigid as the British to do silly so well.
I would think our “none of the above” reaction would be staring blankly ahead, which is a very likely reaction if one tells a bad joke. Then again if one tells a really bad joke you may get confusion or disgust at your pathetic attempt instead. Or if it’s really bad it’ll cross the line again and be so bad it’s good.
I’d side with those who say humor is a specific faculty that was probably selected for and not a side effect of a general increase in intelligence. Anecdotally there are socially intelligent people who have the sense of humor of a rock, and it seems many who are funny enough to make a living from it exhibit their talent at a young age. I don’t think a smart human-like computer would suddenly understand humor unless that was put in.
Are you implying sexual humor is particular to North America? I’m pretty sure this is a human universal.
It’s always been funny to me that no one can decide if Americans are repressed Puritans or completely depraved. Maybe the latter is a reaction to the former. Meeting a visiting foreigner who is shocked at how openly Americans talk or joke about sex vs. what we censor on TV isn’t unheard of.
Well, aside from the whole “computers don’t really understand natural language yet” angle, there’s also the part where we don’t technically understand what “funny” is. There’s theories out there, but none universally accepted and most are too vague to be useful in constructing funny. That is, they can help explain why something is funny but are pretty much useless to somebody trying to create a joke.
Humor research is getting a little more formalized – maybe by the time computers fully work with language, we’ll have a viable theory of funny to feed the system.
Edit: All that said, computerized joke generators could (and probably do) exist now, but you’d have to settle for something that’s essentially creating variations on a predetermined joke rather than building a “new” joke.
There are a lot of variations of humor, and many of them involve very subtle details. A computer could produce some wicked puns for instance, but have no idea how to rate them in terms of their effect on humans. The result would be numerous pointless puns with an occasional gem popping up by accident. Much of humor has to do with the human experience though, and computers aren’t very good at that. In addition, except in written form, they don’t have much of a chance right now. Verbal and physical jokes are much more complicated, and often involve interaction with the audience. Like everything else, one day computers will do it better than we do. But for now, comics will still have jobs. Low paying, degrading jobs. But it still beats working for a living.
My sense is that American culture is, relative to most of the Western world, particularly hung up and conflicted about sex, and therefore more obsessed about it. No cite to support the contention however.