Making AI Work for Hiring

Three critical recommendations

The latest HR.com study suggests we’re at the cusp of significant AI adoption in recruitment. Just over a quarter (28%) of HR professionals have at least a moderate use of AI for hiring today, and more than half (61%) expect to get there within the next two years. Interest in the topic of artificial intelligence may be strong but moving from interest to adoption can be easier said than done.
 

The IBM Smarter Workforce Institute recently completed a research study based on IBM’s internal HR and Talent Acquisition team adoption of AI. The report contains critical insights that can help HR professionals begin their journey toward adoption. 

It’s not enough to accept what comes out of an AI ‘black box’, you need to know how an AI system has arrived at a particular outcome or recommendation. Professional vendors will welcome discussions about how their AI works, and that transparency will reassure your hiring team that they can trust the output. Without an understanding of how AI works, you may find it challenging to show that algorithms are both valid and job-related (potential counters to adverse impact). Using subject matter experts (SMEs) to review data input as well as AI output can also help confirm that there are no instances of adverse impact (where it’s not job-related) in the data used in AI algorithms.
 
Industrial-organizational psychologists are one of the best choices of SMEs to consult with on AI development and deployment. This is primarily because they’ve had formal training in analyzing people data that helps them identify and avoid bias. Other SMEs to consider are HR or legal professionals, data scientists, and project managers.  It is also important to note that it’s helpful to add representatives of groups that may be impacted adversely to the AI deployment team. 

Some people believe that AI will ‘make decisions,’ which is a concerning misconception. A significant theme discovered in the Smarter Workforce Institute study is that companies have a stewardship responsibility to ensure that AI empowers employees. AI can inform choices, but the final decision should rest with people and not machines. Humans feel most empowered when our decision-making autonomy is augmented rather than replaced. AI is only successful if viewed favorably and used effectively. IBM found that AI was viewed most favorably when managers had the option to override AI recommendations when they were not seen as optimal.
 
One of the distinctive traits of AI is that it learns so that when users provide feedback on why a recommendation or suggested action is not optimal, the AI system can improve. Ultimately, this interaction enhances both the quality of AI output and the experience of the user.    

Another insight from the IBM Smarter Workforce study is the criticality of the skills needed to deploy AI successfully.  Unless you’re planning on developing your own AI, it’s the deployment skills that deserve attention. AI expertise is not a requirement, but an analytical approach is advisable. Being able to ask the right questions when troubleshooting, such as where the data came from, who trained the AI application, what were their motives, and is it appropriate to use the data in the way intended, is the key to a successful deployment. Everyone on the implementation team, and ideally the end users, need to be educated consumers of the technology.

Once your deployment team has mastered these necessary skills, there’s almost no excuse not to ‘try out’ AI. You don’t have to go straight for an enterprise-wide deployment, in fact, it’s much better to start small with a pilot designed to address a specific business need. Set clear evaluation criteria, learn from that experience, and expand from there.
 
Read the full report here.
 
The Smarter Workforce Institute hopes to inform and be of service to anyone who is starting an AI recruiting journey through their studies of real teams doing the same. If you’ve already adopted AI in recruitment, we’d love to hear about your experiences to gain and share further insight to help even more businesses start their AI journey.
 
To learn more please visit www.ibm.com/talent-management.