You think the LSAT is a poor indicator of intelligence? Of all the standardized tests I’ve taken, I think it’s probably the best. It’s pure logic with no knowlege required. Lots of problem solving and lots of inference.
I’d say it’s a far better indicator than the SAT, the ACT, the GRE, or the MCAT, anyway.
If the LSAT is a good measure of intelligence, then computers surpassed humans in intellect back in the 1970’s.
The so-called logic puzzles on the LSAT don’t even require first-order logic inferences! They’re “constraint satisfaction problems”, which are not much more difficult than basic arithmetic for a computer to solve. You can even solve them with Excel!
The big difference between those puzzles and basic arithmetic is that you were taught how to solve the arithmetic problems in school.
You certainly don’t think that someone who can do a lot of arithmetic in his head is a genius, do you? But if no one ever taught you multiplication or long division, and you saw someone who figured out how to do it, you’d probably think they were smart.
If you want to measure someone’s intelligence, why not look at how well they play CHESS, something computers cannot do very well. I think there’s a reason why all these games that have been played by humans for hundreds of years are nearly impossible for modern machines to cope with.
The only thing that makes it hard for machines to solve LSAT puzzles is that they don’t understand English very well. From this we can infer that any decent poet is much more brilliant than the law student who got a perfect LSAT score!
But for those who are interested, there was a natural language project by Chris Manning and others at Stanford that attempted to make a program that could read and solve LSAT puzzles in English.