Presented By O'Reilly and Cloudera
Make Data Work
31 May–1 June 2016: Training
1 June–3 June 2016: Conference
London, UK

Removing human bias from the interview process

Gary Willis (ASI)
16:35–17:15 Friday, 3/06/2016
Data science & advanced analytics
Location: Capital Suite 8/9 Level: Intermediate

It is widely recognized that human bias has a significant impact on any company interview and assessment process. Traditional recruitment techniques make it virtually impossible to assess and adjust for human biases. The rising popularity of recorded interviews in the recruitment sector presents a potentially game-changing way to address and remove unconscious biases from the interview process.

Gary Willis explores how a quality assurance module that monitors human reviewers can correct their responses for biases. This so-called “blackbox” receives anonymized data on how reviewers score candidates’ responses to questions and performs bias adjustment calculations in real time. One of the key innovative ideas fundamental to the process is the ability to show different reviewers the same candidate.

This approach allows for the data scientists to quickly and effectively iterate their models in isolation and adjust candidate scores in real time—one of the most innovative data-driven recruitment solutions in the world. Gary offers an overview of the technical challenges associated with this approach and some of the insights gained from using it successfully for multiple recruitment campaigns.

Photo of Gary Willis

Gary Willis


Gary Willis is a data scientist at ASI with a diverse background in applying machine-learning techniques to commercial data science problems. Gary holds a PhD in statistical physics; his research looked at Markov Chain Monte Carlo simulations of complex systems.