Scientists Invented a Simple New Test for Creativity, and You Can Try It Out

It seems like there's an interesting bias in the test methodology.

If you are a person who prefers simple, common or short words, the test is much harder because short words tend to have many distinct meanings.

When you cross correlate a lot of short words there will inevitably be semantic overlap.

Longer, more specialised words have greater semantic distance.

Therefore not clear whether this measures creativity or propensity to use longer / less common words. This will tend to correlate with "years of formal education".

It would be interesting to see if this is measuring anything different than a vocabulary test + the ability to understand the instructions.

One option to balance this out would be to "weight" words by their frequency to reflect that finding disjoint common words is harder.

Here's the thing. Creativity is measured by a subjective human mind. If you can simulate the mind's current state using AI (with relevant inputs), you can probably measure creativity.

For some reason, we humans glorify certain things like creativity, attributing those to some insurmountable human ability. As a scientist, I don't find anything extraordinary in this. Just a simple ability to create. After all, who is is the judge? Ourselves!

For anyone interested, what the study does is, asks you to name 10 different nouns, and then searches for the semantic distance (how often they are not used together) between them to give you a score.

So I just fed it a list of computer generated random nouns with 2 syllables each and got a 95th percentile.

Score of 90. I doubt this really invokes creativity (in my case). I just start making a vowel sound in my head and then finish it with a word, so its more sound oriented than based on association.