Model Sensitivity

Aidan shares the surprising sensitivity of models to human-collected data, highlighting the critical impact of even a few errors on model direction. Lukas expresses astonishment at the delicate nature of these models, shedding light on the challenges beyond just training them.