In recent years behavioural science has made great advances in the reliability of facial recognition techniques. Current evidence suggests algorithmic facial identification systems are now more reliable than human attempts to recognise faces. This leads some to have confidence in the capacity of computational systems to read human emotions in a way that will be more reliable than interpersonal interpretation. To those optimists, I present the really hard problem facing data driven behavioural science (HT Shit Academics Say):
For a more serious attempt to make the same point, see this post from a few weeks ago.
Categories: Digital Sociology