Hacker News new | ask | show | jobs
by Vecr 801 days ago
Can't you train a predictor on skull shape/bone length and IQ and go from there? Yeah there's no causation there, but you should be able to give a percentage that someone's IQ is above a certain number depending on high-rez MRI scan input of skull and bone length measurements.
2 comments

You can train a predictor on anything: that's how the famous cases of 'racist AI' came about. The point is, you can't establish a causal relationship without accounting for confounders. If you're starting with a question established by racists, with 'race' and 'IQ' as your two poles, and human populations as your data, good luck managing that.
Or make asset worth a predictor.