Guest blog by Roisin McCarthy Manager at Datatech Analytics and Payal Jain, Managing Director at JCURV who will be speaking at this year’s Housing Data and Analytics Summit.
Building a successful team is about more than finding a group of people with the right mix of professional skills… think about your workplace and those you work closely with. If you noticed that everyone is the same gender and ethnicity, ask yourself why that is? For those of you who are surrounded by diversity, your work place is exceptional… also ask yourself: how come?
As the power of data is unlocked across different industries, including recruitment, data professionals must consider how historical data can influence the process and lead to selection bias. When you have AI, the bias can be even more pronounced, not because algorithms are intentionally evil, but because they lack awareness required to identify their own biases. AI are often programmed to learn from historical data.
Computer science has a concept known as GIGO (Garbage In Garbage Out); quality results depend on quality data. In the case of AI, if biased historical data goes in… diversity goes out. To use an abstract example, if your organisation historically hires hares in accounting, that information informs your recruitment AI algorithm. The next time you are hiring an accountant, even when there is a tortoise with a more reliable track record, the AI will select the hare. In the real world, these biases affect every aspect of our lives from the amount banks will lend us, to our position in society.
At this year’s Housing Data and Analytics Summit Payal Jain (MD JCURV & Chair of Women in Data) and Roisin McCarthy (Co-Founder WiD, Datatech Analytics) will be stepping up to the podium to deliver a presentation on ‘How to build a successful data team’ which will include important insights on bias and how to combat it.
The Housing Data and Analytics Summit is taking place on 1 October at Millennium Point, Birmingham. To find out more and secure your place click here.
No Related Items
The November issue of our Data Analysis Bulletin is now available
Catch up on the October edition of our newsletter
We are pleased to share the findings from HouseMark’s online resident survey, completed by close to 8,000 people