Summary:
- Jokes are hard to explain and highlight how we cannot explain creativity.
- Many engineers and scientists have tried to achieve creativity using formulas.
- Creativity might be something different than formulas.
Background:
David Deutsch thinks that achieving human level intelligence with computers is a matter of understanding the essence of creativity (The Beginning of Infinity). A joke is an interesting example of creativity because humour is difficult to explain.
Is joking “Intelligence”?
One type of joke involves taking something from one context and replacing it with something from another context that is only related in a limited – and often absurd – way.
Girl (holding up an an apple): Mammy, would you like to try this vaccine?
Mother: That’s not a vaccine young lady. It’s an apple.
Girl: Sorry, I thought I’d give it a shot.
Sample joke.
I don’t think I can perfectly explain why this joke is funny, but let me try anyways:
Attempt #1: The joke is funny because the word “shot” is a pun. This is a bad explanation because it now kicks the can to explaining the word “pun”.
Attempt #2: A joke is funny because it explains something non-obvious. People appreciate explanations, particularly non-obvious explanations, and react by laughing. In this case, it is non-obvious why the girl would call an apple a vaccine. The explanation is that calling an apple a vaccine does make sense if it is a reminder of the broader context of the COVID outbreak the girl and her mother are living through. This metaphor is further engrained by the pun, “shot”, which ties together the attempt at a joke and the vaccine. I think this is a bit better as an explanation but it is also a bit circular because I’m saying that the joke is partly funny because it is pointing out that the setup line is a joke.
Ok, I’ve run out of more ideas, but comment below if you have some.
The temptation with creativity is to find a formula
Computers are good at implementing formulas. However, formulas/models seem to be human. Even if the computer is “machine learning” like Alpha Go for Chess, there were people who developed the model.
If we want computers to come up with their own formulas, how would we do that? The temptation again is to come up with a formula that allows computers to come up with formulas.
This leads to the question of whether artificial intelligence could be achieved simply by having sufficient layers of formulas. This is what it seems humans have been trying to do roughly since Alan Turing defined a test of whether an AI is real back around the second world war – use more formulas with more data and more computing power.
Another possibility is that creativity is a separate thing to formulas. The End.