Over a million developers have joined DZone.
{{announcement.body}}
{{announcement.title}}

Can AI Do a Better Job of Recruiting Directors?

DZone's Guide to

Can AI Do a Better Job of Recruiting Directors?

A recent study has used artificial intelligence to look at the characteristics that make a potential company director effective and popular.

· AI Zone ·
Free Resource

Start coding something amazing with the IBM library of open source AI code patterns.  Content provided by IBM.

Various biases have been well known to exist in recruitment for some time, while securing a diverse boardroom has also been shown to boost the performance of organizations. Alas, as CEOs typically control the recruitment of directors, it's very easy for the board to be a mirror image of the CEO.

Might AI technology do a better job? That was the question posed by a recent study into whether machine learning can select directors more effectively than current practices.

For Better or Worse?

First things first: What exactly makes a director effective? With many of their actions done in private, the researchers used the votes each director receives in annual shareholder re-elections as a proxy of their effectiveness. These votes hopefully reflect all of the publicly available information on each director's performance.

This method also creates a challenge, however, in that it only includes those directors who were selected rather than those who were considered but ultimately not hired. Predicting future performance of directors will require both groups to be included. The researchers overcome this challenge by having a wide pool of directors who had recently accepted board positions at smaller companies. As we can assume these directors would accept a position at a larger nearby firm, they are useful in compiling the pool of potential directors for each position.

The algorithm was trained on data of directorships at publicly traded US companies between 2000 and 2011. The algorithm then tested itself on a second dataset of fresh directors who joined their firms between 2012 and 2014.

Proving Popular

The algorithm proved adroit at predicting the popularity of directors with shareholders. When the algorithm predicted a director would be unpopular, this did, in fact, materialize, while the inverse also emerged, with those directors predicted to do well outperforming rival candidates.

By highlighting the directors who performed best, the researchers were then able to examine those directors to see if any distinguishing features stood out. This analysis revealed that the most commonly selected directors were well-networked men with a lot of previous board experience and who come from a financial background. In other words, it's a pretty homogeneous bunch.

Suffice to say, the project isn't yet able to directly ensure that boards are more diverse than they currently are, but hopefully, it will at least shed some light on the lack of diversity in boards today and therefore encourage companies to do a better job.

They do, however, point to a possible future whereby algorithms become more sophisticated and capable of doing a more fair and effective job at recruiting than humans can. This could be especially so if given access to private data rather than relying purely on publicly available data to train on. Of course, the question then is whether CEOs would want an algorithm to select their board for them. That, perhaps, is a tougher nut to crack.

Start coding something amazing with the IBM library of open source AI code patterns.  Content provided by IBM.

Topics:
ai ,algorithm ,deep learning ,training data

Published at DZone with permission of

Opinions expressed by DZone contributors are their own.

{{ parent.title || parent.header.title}}

{{ parent.tldr }}

{{ parent.urlSource.name }}