Episode Summary: How do you know if you’ve made the right decision for a hire? Often, employers go off gut instinct and make a decision retrospectively, but it turns out AI might be able to help out in human resource management through shedding light on best hiring decisions. In this episode, Pasha Roberts, chief scientist at Talent Analytics, tells us about how his company is working on helping companies make better decisions before they hire by applying machine learning and artificial intelligence to various data points on a given applicant, including information from aptitude tests that may help predict not only performance but retention.
Expertise: Data science and predictive analytics of workforce behavior
Brief Recognition: Pasha is an accomplished entrepreneur, having founded 2 other companies in addition to Talent Analytics. After winning MIT’s Business Plan contest, he founded WebLine Communications Corporation, which was eventually acquired by Cisco. He also founded Lineplot Productions, a financial visualization/animation service company, where he directed Silver Circle, an independent film. Pasha is also a frequent speaker, having presented at major predictive analytics events including INFORMS and Predictive Analytics World, and writes extensively on solving complex workforce challenges.
Pasha holds a Master of Science Degree from the MIT Sloan School of Management in Financial Engineering and dual Bachelors degrees in Economics and Russian Studies from The College of William and Mary. His thesis at MIT prototyped the application of advanced 3D graphics to massive financial “tick” datasets.
Current Affiliations: Chief Scientist and Co-founder of Talent Analytics
(1:25) Today at present, what are the AI applications used to predict job performance?
(3:29) When you say ‘is this kind of person’, what is ‘kind of person’ to your company, what information can you look at…what are we able to pull into the system to make those calls?
(7:31) Are you able to pull (CV) into your models when you’re making these calls? I would understand how something as quantitative as an assessment could be shoved into a system, but can we really say “spent 2 years as ‘such and such’”…how is that being able to inform your ML models?
(9:37) Are we looking here at high-volume jobs that are very measurable (in order to build a hiring model)?
(12:27) It seems to me that maybe…different companies, this has to be more bespoke; maybe you’re not able to take what you learned at call center A and automatically plug it in…are there trends that you’ve been able to pool…or are we really looking at training a pretty unique model per business?
(14:58) Where do you see the biggest changes in AI in HR over the next 5 to 10 years?
(19:15) Do you see any particular applications that are so tangible and low-hanging fruit…that they’re likely to be what is adopted widely (in this area)?
Related Emerj Interviews/Articles: