| 1 |
Labour market discrimination and biases in human judgement and Artificial Intelligence ...
|
|
|
|
Abstract:
Racial discrimination in labour market experiments demonstrate a preference for names commonly associated with White candidates over other ethnicities, such as African American (Bertrand & Mullainathan, 2004). These experiments typically involve near identical resumes being sent in field trials to job advertisements, but with different names attached. These field experiments demonstrate a bias in human decision-making concerning racial preferences that has been replicated numerous times (Nunley, Pugh, Romero and Seals, 2015). At the same time, there have been advances in computational linguistics where researchers have demonstrated multiple instances of machine learning and artificial intelligence systems demonstrate learnt "biases" predicated on written texts. In these outputs, the systems generate similar biases as observed within human judgements, for example, predictive policy systems have demonstrated the potential to get caught in positive feedback loops to over-police certain areas with higher ...
|
|
Keyword:
Bias; Biases; Computational linguistics; Discrimination; Economics; FOS Psychology; Human judgements; Judgement and decision-making; Labour markets; Machine Learning; Psychology; Social and Behavioral Sciences
|
|
URL: https://osf.io/dneup/
https://dx.doi.org/10.17605/osf.io/dneup
|
|
BASE
|
|
Hide details
|
|
|
|
|
