How Artificial Intelligence Is Changing Talent Acquisition

AI for recruiting is on everyone’s mind these days with a lot of talk on how it’s going to transform recruiting. Artificial intelligence for recruiting is the next generation of software designed to improve or automate some part of the recruiting workflow.

Interest in AI for recruiting has been sparked by three major trends:

  1. The improving economy: The recent economic gains have created a candidate-driven market that’s made competing for talent tougher than ever. This competition will only continue to increase – 56% talent acquisition leaders surveyed by LinkedIn believe their hiring volume will grow in 2017.
  2. The need for better technology: Although hiring is predicted to increase, 66% of talent acquisition leaders state their recruiting teams will stay the same size or even shrink. This means time-constrained recruiters need better tools to effectively streamline or automate a part of their workflow, ideally for tasks that are the most time-consuming.
  3. The advancements in data analytics: As technology becomes fast and cost-effective enough to collect and analyze vast quantities of data, talent acquisition leaders are increasingly asking their recruiting teams to demonstrate data-based quality of hire metrics such as new hires’ performance and turnover.

The growing popularity of AI for recruiting represents exciting opportunities for recruiters to enhance their capabilities but there’s also a lot of confusion about how to best leverage it.

To help you make sense of it all, here are the three most promising applications for AI for recruiting.

Application #1: AI for candidate sourcing

Candidate sourcing is still a major recruiting challenge: a recent survey found 46% of talent acquisition leaders say their recruiting teams struggle with attracting qualified candidates.

AI for candidate sourcing is technology that searches for data people leave online (e.g., resumes, professional portfolios, or social media profiles) to find passive candidates that match your job requirements.

This type of AI for recruiting streamlines the sourcing process because it can simultaneously search through multiple sources of candidates for you. This replaces the need to manually search them yourself and potentially saves you hours per req. The time you save sourcing can be spent attracting, pre-qualifying, and interviewing the strongest candidates instead.

Application #2: AI for candidate screening

When 75-88% of the resumes you receive are unqualified, it’s easy to see why resume screening is the most frustrating and time-consuming part of recruiting. For high-volume recruitment such as retail and customer service roles, most recruiting teams just don’t have the time to manually screen the hundreds to thousands of resumes they receive per open role.

AI for screening is designed to automate the resume screening process. This type of intelligent screening software adds functionality to the ATS by using post-hire data such as performance and turnover to make hiring recommendations for new applicants.

It makes these recommendations by applying the information it learned about existing employees’ experience, skills, and other qualifications to automatically screen and grade new candidates. This type of technology can also enrich resumes by using public data sources about previous employers and candidates’ social media profiles.

AI for resume screening automates a low-value, repetitive task and allows recruiters to re-focus their time on higher value priorities such as talking and engaging with candidates to assess their fit.

Application #3: AI for candidate matching

Candidate matching can be an even bigger challenge than sourcing: 52% of recruiters say the hardest part of their job is identifying the right candidates from a large applicant pool.

AI for candidate matching uses an algorithm to identify the strongest matches for your open req. Matching algorithms analyze multiple sources of data such as candidates’ personality traits, skills, and salary preferences to automatically assess candidates against the job requirements.

For example, a LinkedIn job posting ranks candidates by matching the skills on your job description to applicants’ skills on their LinkedIn profiles. Talent marketplaces use matching algorithms to match their community of candidates to open roles. These talent marketplaces usually cater to specific candidate skill sets such as software development or sales.

AI for matching is used to identify the most qualified candidates from those who have opted-in and are either actively looking for a new role or are very open to a new opportunity. This means recruiters don’t need to waste time trying to attract passive candidates who just aren’t interested in a new role.

Article Continues Below

For a different perspective on the power of artificial intelligence to match candidates to jobs see “Despite What You Read or Hear, Sourcing Is Alive and Well Indeed.”

AI and the future of recruiting

Experts are predicting AI for recruiting will transform the recruiter role. As low-value, time-consuming recruiting tasks become streamlined and automated through AI technology, the recruiter role has the potential to become more strategic.

Recruiters who understand how AI will augment their capabilities will reap the benefits of increased efficiency through the dozens of hours saved per hire on sourcing, resume screening, and candidate matching.

AI for recruiting promises to free up recruiters’ time to engage with candidates to determine fit and pinpoint candidates’ needs and desire to persuade them to take the role. It holds the potential to empower them to partner with hiring managers and talent acquisition leaders to plan out proactive hiring initiatives based on future growth and revenue rather than reactive back-filling.

Recruiters who figure out how to best leverage this new technology will be rewarded with improved KPIs such as higher quality of hire and lower turnover.

  • steventhunt

    Why are we calling these methods “artificial intelligence” instead of just calling them complex, non-linear mathematical modeling? The methods in this blog are useful for sure, but they aren’t what I’d call artificial intelligence. Most of them are forms of mathematical pattern recognition and/or closed loop predictive modeling and validation methods. They involve complex math formulas that operate in a very different matter than the way the human brain solves problems. They bear little in common with actual human cognitive processes and calling them artifical intelligence is a bit like calling a Jolly Rancher candy an “artifical fruit”.
    Describing math formulas using terms like “artificial intelligence” might make these things sound super futuristic but I believe it can also create distrust and confusion. We should talk about how these things augment and complement human intelligence, not how they replace it. Otherwise we are inviting public backlash against their use since I suspect most people don’t want their futures totally controlled by machines (at least not those of of us who grew up watching the Terminator movies).

    • Ji-A Min

      Hi Steven: These technologies fall under the umbrella of AI because they often use a machine learning algorithm to improve their pattern matching.

      I completely agree that the emphasis should be on how AI will augment human abilities and I’ve re-iterated that point through out including my conclusions.

      • steventhunt

        Ji-A, To be clear, I like what you’ve written and how you presented it. I’m just reacting to the use of the term artificial intelligence (AI). I fear it has become a marketing buzzword.

        Originally AI programming was designed to mimic how the brain was believed to work. The goal was to model human cognitive processes using computers and see if we could get computers to “think like people”.

        Most modern machine learning algorithms that use things like non-linear multivariate
        regression techniques and/or bayesian statistical methods most definitely do NOT mimic human cognitive processes. To the contrary, the value of these algorithms is they process information much differently from how humans think. These innovations are not about creating “artificial intelligence” but about using mathematical calculations to process information in a way humans never would or could. So why not just call them what they are: complex mathematical formulas.

        • Rosario

          A lot of people in recruitment are massive fans of buzzwords. Especially when working in internally, and you’re held to account over a potential failure to attract the right calibre of staff. I’ve sat in meetings during which the heads of internal recruitment throw all the buzzwords around to make the whole talent attraction thing sound very complicated. And generally when something sounds complicated, you’re less likely to get asked questions. Its an effective tool for some.

    • Mickelodian

      Well as someone in the field maybe the process was not defined in the article in terms of the general machine learning process. As fas as I see here and I nonhiring expert but the goal is tonmatch the best applicants to say 100 people currently in a similar role and performing at optimum. So the machine uses these 100 folks as the target or output objective and it sees the resumes entered the first time as the training data. The objective of the software would be to use the resumes submitted by say 10,000 applicants for these 100 vacancies as a test.. It can later apply the learning to new roles and new applicants.

      That’s not obviously a deterministic problem. So machine learning would be a good solution.

      But you are right in that brute force pattern matching and hard coded rules (if this then that) could also be used. However you’d need to run it every time and It would not be as effective.

      Plus if an agency or HR department have the data or a way of describing to an developer what the data looks like then this is easier to code using a simple AI or ML app like tensorflow or Torch.

      I would say all thoughts of deterministic solutions are currently evaporating in the HR game to be honest.

  • Rosario

    The percentages you’ve listed – What region are they attributed to exactly? 46%, 56%, 66% – did internal recruiters within the UK / EU market also get asked for their thoughts? I would suggest that more of these questions are actually answered by hands on internal recruiters rather than so called ‘leaders’. I say that because I’ve worked in environments with multiple so called ‘leaders’ who’s input and insight into actual recruitment is very little, whilst they tend to be very good at putting together pie charts and graphs.

  • Job Lagao
  • Isaac Marks

    Here is the problem, AI or not AI , when there algorithms identify a candidate they identify a pice of paper in your ATS which matches your job description. I give them that. But at the end your algorithms interact with paper not candidates and candidates embellish, or misstate their experience. So the problem continues because this is a data issue: garbage in ,garbage out. This AI matching technology should be coupled with additional recognition technologies anf find a way to interact with the candidates directly once identified by the AI system. Then it makes sense, otherwise this will be one of many HR TECH toys appealing to enterprises until they see the value for the huge price they charge does not correalate.

  • Bas van de Haterd

    Although it might be implicit in screening / matching, I’m missing one big aspect. AI might actually be able to tell us the potential of a candidate. Seldom have I seen a linear career path. And everybody is always so surprised about the moves they make themselves, while usually many others do so as well. Based on for example Linkedin data it should be possible to predict someone’s potential for future roles (no linear, so not junior to medior to senior) and hire accordingly. And also, based on that potential maybe suggest certain test (character) if the person would have the skills or would be able to obtain the skills based on his or her ability.