Study predicts criminality by analysing facial features

Occurred: November 2016

Can you improve this page?
Share your insights with us

Shanghai Jiao Tong University researchers Xiaolin Wu and Xi Zhang say they created a system which is able to accurately predict whether someone is a criminal by analysing a few of their facial features. The claim was widely panned, and resulted in accusations of poor ethics, physiognomy, and pseudo-science.

The researchers took ID photographs of 1856 Chinese men between the ages of 18 and 55 with no facial hair, scars or other markings, half of which were criminals. They then used 90 percent of these images to train a convolutional neural network to recognise the difference and tested the neural net on the remaining 10 percent of the images. The result: the neural network was correctly able to identify criminals and noncriminals with an accuracy of 89.5 percent.

Critics took issue with the research supposition, approach, and methodology, especially with regard to potential data bias, and expressed concerns about how data of this kind could be misused and abused, including in an authoritarian Chinese context.

The researchers responded by saying 'Our work is only intended for pure academic discussions; how it has become a media consumption is a total surprise to us. Although in agreement with our critics on the need and importance of policing AI research for the general good of the society, we are deeply baffled by the ways some of them mispresented our work, in particular the motive and objective of our research.'

Operator: Shanghai Jiao Tong University
Developer: Xiaolin Wu; Xi Zhang
Country: China
Sector: Research/academia
Purpose: Predict criminality
Technology: Computer vision; Neural network; Deep learning; Machine learning
Issue: Accuracy/reliability; Dual/multi-use; Ethics