LLMs are solving MCAT, the bar test, SAT etc like they’re nothing. At this point their performance is super human. However they’ll often trip on super simple common sense questions, they’ll struggle with creative thinking.
Is this literally proof that standard tests are not a good measure of intelligence?
such tests are not standardized tests of intelligence, they are standardized tests of specific-competencies.
Thomas Armstrong’s got a book “7 Kinds of Smart, revised”, on 9 intelligences ( he kept the same title, but added 2 more ).
Social/relational intelligence was not included in IQ because it is one that girls have, but us guys tend to not have, so the men who devised IQ … just never considered it to have any validity/significance.
Just as it is much easier to make a ML that can operate a commuter-train fuel-efficiently, than it is to get a human, with general function, to compete at that super-specialized task, each specialized-competency-test is going to become owned by some AI.
Full-self-driving being the possible exception, simply because there are waaaaay too many variables, & general competence seems to be required for that ( people deliberately driving into AI-managed vehicles, people throwing footballs at AI-managed vehicles, etc, it’s lunacy to think that AI’s going to get that kind of nonsense perfect.
I’d settle for 25% better-than-us. )
Just because an AI can do aviation-navigation more-perfectly than I can, doesn’t mean that the test should be taken off potential-pilots, though:
Full-electrical-system-failures do happen in aviation.
Carrington-event level of jamming is possible, in-flight.
-
Intelligence is “climbing the ladder efficiently”.
-
Wisdom is knowing when you’re climbing the wrong ladder, & figuring-out how to discover which ladder you’re supposed to be climbing.
Would you remove competence-at-soccer tests for pro sports-teams?
“Oh, James Windermere’s an excellent athlete to add to our soccer-club! Look at his triathelon ratings!”…
… “but he doesn’t even understand soccer??”
… “he doesn’t need to: we got rid of that requirement, because AI got better than humans, so we don’t need it anymore”.
idiotic, right?
It doesn’t matter if an AI is better than a human at a particular competency:
if a kind-of-work requires that competency, then test the human for it.
-
Intelligence cannot be measured. It’s a reification fallacy. Inelegance is colloquial and subjective.
If I told you that I had an instrument that could objectively measure beauty, you’d see the problem right away.
But intelligence is the capacity to solve problems. If you can solve problems quickly, you are by definition intelligent.
the ability to apply knowledge to manipulate one’s environment or to think abstractly as measured by objective criteria (such as tests)
https://www.merriam-webster.com/dictionary/intelligence
It can be measured by objective tests. It’s not subjective like beauty or humor.
The problem with AI doing these tests is that it has seen and memorized all the previous questions and answers. Many of the tests mentioned are not tests of reasoning, but recall: the bar exam, for example.
If any random person studied every previous question and answer, they would do well too. No one would be amazed that an answer key knew all the answers.
This is a semantic argument.
Have you never felt smarter or dumber depending on the situation? If so, did your ability to think abstractly, apply knowledge, or manipulate your environment change? Intelligence is subjective (and colloquial) like beauty and humor.