Yes, sort of. Generally you can measure the pass rate on a benchmark given a fixed compute budget. A sufficiently smart model can hit a high pass rate with fewer tokens/compute. Check out the cost efficiency on https://artificialanalysis.ai/ (say this posted here the other day, pretty neat charts!)
It's much easier to measure a language model's intelligence than a human's because you can take as many samples as you want without affecting its knowledge. And we do measure human intelligence.
How do I reliably measure 1 unit of intelligence?
In pelicans, obviously
Isn't the outcome / solution for a given task non-deterministic? So can we reliably measure that?
Yes, sort of. Generally you can measure the pass rate on a benchmark given a fixed compute budget. A sufficiently smart model can hit a high pass rate with fewer tokens/compute. Check out the cost efficiency on https://artificialanalysis.ai/ (say this posted here the other day, pretty neat charts!)
Statistically. Do many trials and measure how often it succeeds/fails.
This is the only correct take. The only metric that matters is cost per desired outcome.
Repetition and statistics, if you have $1000++ you didn't need anyway.
It's much easier to measure a language model's intelligence than a human's because you can take as many samples as you want without affecting its knowledge. And we do measure human intelligence.