I just received the final results from our annual Quality of Life Survey (longitudinal data available on many questions now from 1985-2008!) I'm also finalizing the questions for our Race Relations Progress Report, which is a little complicated this year as we expand the number of populations within which we hope to receive statistically significant results.
So this note was timely, from the linguistically-obsessed folks at Language Log. Just a reminder of some of the pitfalls in constructing and conducting surveys.
Take a look!
Jagged intelligence demonstrated with a puzzle
-
Large language models are heavily dependent on training data to determine
what comes…
*Tags:* Aatish Bhatia, puzzle, training
19 hours ago

0 comments:
Post a Comment