Humanitarian aid guided by satellite data may harm marginalised groups

By Edd Gent

A outer orbits Earth supra India

Shutterstock / Cristi Matei

Satellite information tin assistance policy-makers rapidly place areas of the satellite successful request of assistance and development, but probe shows it tin besides incorporate bias against marginalised groups, perchance compromising argumentation goals.

Machine-learning systems that scan outer images for indicators of poorness oregon catastrophe harm are becoming a fashionable instrumentality for assessing humanitarian and improvement needs. But researchers astatine the German Aerospace Center successful Cologne accidental small attraction is being paid to imaginable biases built into this data.

The radical trained a model to …