ML intern beats Astra at ideating research solutions, engineer observes

andrew_n_carr · x · 2026-09-15

The author compares ML interns they've mentored with Astra: interns consistently had impressive ideas for ML solutions, while Astra was merely "fine" and nowhere near as good at ideating. A first-hand data point on current models' weakness in creative research ideation versus humans.

Original post →

More from Models

Models channel →