AI for Recruiting 101 [Video]

Devin Foster

This article recaps the March 18 episode of Talent Experience Live. Kumar Ananthanarayana, Director of Product Management, and Ilya Goldin, Principle Data Scientist for Phenom, joined us to break down the fundamentals of AI and its potential to drive incredible outcomes for recruiting. 
 

At Phenom, our AI technology experts field questions from TA teams across the globe related to using artificial intelligence (AI) for recruiting. What are AI’s best use cases? What are limitations of AI? Can AI be used to improve diversity and inclusion efforts? And – most pressing – will using AI create evil robots that take over the world?
 

Dive in with us for a deeper understanding of how to best apply AI in recruiting. On last week’s episode of Talent Experience Live, Phenoms Kumar Ananthanarayana and Illya Goldin explained how AI can help you personalize the candidate experience, find the right talent faster, and enhance initiatives like D&I and internal mobility. (Also covered: What do Netflix and pizza have to do with AI in recruiting?) Watch it here — and catch highlights below!
 

 

What are the top use cases for AI for recruiting?
 

AI-powered tools and automation help recruiters work more efficiently, especially with high-volume, repetitive tasks like screening, sourcing and scheduling. Not only do this check a few tasks off their to-do lists, it also gives them time back to develop relationships and nurture key candidates.
 


So AI drives efficiency, but what about its role in helping solve complicated problems? 
 

AI helps bring clarity to decision-making that’s complex and involves too many unpredictable data points to represent in a small rule-based presentation, Goldin said. (Rule-based AI systems use if/then references to build knowledge around decision making.)
 

To use Netflix as an analogy, consider what would happen if viewing recommendations were limited by an if/then rule that watching Star Wars movies only generated recommendations for Harry Potter movies. Viewers would get bored pretty quickly, and it fails to take into account other preferences or new movies that come along. The same goes for technology that recommends take-out options – just because you like pizza doesn’t mean you want Italian food every single night. 
 

In TA, system programmers can’t proactively create rules that take into account specific candidate career goals, nuances of experience and skill level, and job preferences. “When you have to make a probabilistic guess about circumstances you’ve never seen, that’s a defining characteristic of problems to which we usefully try to apply AI,” Goldin said. 
 

When you think about it, there are a lot of areas like this in HR. For example, we know that it’s important to personalize job recommendations on career sites – but we need to do this for candidates we’ve never met, with new job roles constantly posted. AI-driven automation can generate recommendations based on the information that is available, and improve the accuracy of recommendations over time. 
 

Chatbots and intelligent search capabilities are other AI-powered tools that can assist in this experience, helping job seekers find what they’re looking for in a flash.

 

How can we leverage AI for talent management within the organization?
 

AI is powering better talent retention and employee development, Ananthanarayana said. And systems have gotten more sophisticated: “It’s not just about looking at who the employee is from a profile perspective and recommending jobs. It’s gone a little further, looking at data points like performance, industry average tenures, readiness to move and other external characteristics that can be part of the recommendation story.”
 


TA and HR teams are using AI-driven platforms to provide visibility for employees regarding career path opportunities, including:
 

  • Learning and upskilling 
  • Gig work recommendations 
  • Mentoring programs 


“Employees leave a company because they don’t see a path forward,” Kumar said. TA and HR leaders can use AI-based data analysis to identify patterns of fast career growth within the organization, he explained. “Showcasing that to employees is a huge advantage, and leg up for enterprises today.”

 

Sorting out the acronyms: How Is AI different from ML?
 

It’s easy to be confused by the array of acronyms you’ll encounter in the AI world. Machine learning (ML) in particular goes hand-in-hand with AI, but it’s important to know the difference.
 


AI is the umbrella term for the technology, while ML is a subcategory. ML teaches a system by showing it examples. Then, human involvement is needed to test whether the machine can learn from and apply that rule or pattern to a new context.
 

Another acronym recruiters might hear is NLP, or natural language processing. NLP interprets patterns of human speech or text, enabling chatbot and search capabilities. This is what helps a chatbot or even search bar understand the information you’re providing it with, such as questions someone is asking or jobs they’re looking for.
 

What are some limitations of AI to watch out for?
 

As promising as AI technology is for recruiting, it does pose some risks. (Fortunately, evil robots taking over the world is not one we have to worry about, Goldin said.)
 


As far as realistic concerns go, TA professionals and recruiters should always aim to understand the capabilities and outcomes of AI — this is what will drive the appropriate usage of AI. Here are a couple of key considerations:
 

1. Inaccurate patterns that work against demographic groups

The potential for systemic patterns of inaccuracy that could create unfair bias toward certain candidate or employee groups concerns Goldin the most. Inaccuracies will always exist, he pointed out. But when inaccuracies develop into patterns that influence AI decisions, that’s when problematic outcomes occur. 
 

2. AI pitfalls

According to Ananthanarayana, key considerations can help prevent negative outcomes related to AI: 
 

  • Do we have enough quality data to deploy AI in decision-making? 
  • Do we have enough feedback loops that enable a human to override AI decisions? 


If the answer to either of these is “no,” proceed with caution. 
 


For example, a recruiter may want to prioritize candidates based on AI-calculated scores. The system needs to have enough past training data that’s bias-free, and recruiters need to be able use their own judgment and input feedback to override AI decisions. 
 

Most importantly, recruiters and hiring managers should base hiring decisions more on their human interactions with candidates rather than on fit scoring.

 

How can AI improve Diversity, Equity & Inclusion?
 

Like in other areas, AI’s role in diversity, equity & inclusion (DE&I) is best viewed as supportive rather than definitive: AI won’t solve diversity issues, but it can help create awareness, Ananthanarayana said.
 

AI can predict underrepresented segments within the workforce, and provide insights to help recruiters address those gaps. Here are a couple of examples that outline how Phenom uses AI to help organizations strengthen DE&I:
 

  • Candidate experience: With AI-driven personalization, recruiters can recommend diverse talent communities for candidates to join. (This also builds diversity in the talent pool for sourcing.)
  • Employee experience: AI personalization techniques can match employees with diversity groups and ERGs based on interests and background. 
     


Fine-tuning AI outcomes to start recommending diverse candidates
 

Phenom clients often ask about AI’s role and ability to increase the diversity of interview groups, Ananthanarayana and Goldin said. The technology itself is objective in that it scores and ranks interview candidates in a uniform way. But from a diversity standpoint, AI’s objectivity is limited to the organization’s definition of an underrepresented class, they point out. 

 

What is AI’s role in succession planning and workforce development?
 

AI-driven technology helps provide a more complete picture so that managers and leaders can make stronger succession planning decisions – and avoid backfilling risks: AI can track and analyze engagement patterns, industry data, and attrition risks. It also can help managers identify employees who may need help, and make informed recommendations for a suitable growth plan.
 


“If you think of succession planning, it has always been a once-a-year thing. Managers look at all the people who report to them, and figure out who their successors could be … but the challenge with that is they often miss out on certain key signals, certain key behaviors of employees,” Ananthanarayana said. 
 

Relatedly, AI insights can give HR and TA leaders an edge on workforce planning. AI can uncover existing gaps and recommend the skills needed to close those gaps. 
 

With data inputs regarding the organization and industry, as well as employee behavior patterns, AI can help leaders anticipate future hiring needs, and plan ahead for the right skill mix. 

 

What are some future applications of AI in recruiting?
 


AI is constantly evolving and expanding into new areas that can benefit HR and TA professionals. According to Ananthanarayana and Goldin, the most promising future applications of AI in recruiting include: 
 

Video screening: AI can analyze speech in candidate videos, the outcomes of which can be used as a screening tool for application. Going even further, some are exploring the use of AI for facial recognition and sentiment analysis, although using these insights is tricky from an outcomes perspective.
 

Mentoring: AI can help identify employees who need help developing certain skills or career path aspirations, and match them with mentoring employees who have strong competencies in those areas.
 


Sign up to get notified about future episodes of The Talent Experience Show! Catch us on LinkedInYouTubeTwitter, and Facebook every Thursday at noon ET to get the latest in recruiting, talent acquisition, talent management, and HR tech.