Fig. 2. Strategic Impact Areas of AI in Recruitment
In addition to the strategic advantages, AI in recruiting can also help recruiters achieve tactical recruiting goals as well. Remember how we said modern recruiting is about creating bespoke experiences? AI in recruiting can help organizations with end-to-end candidate experience management as well.
Here are 5 Critical Recruiting Tasks that AI can Automate Intelligently:
- Candidate sourcing: Candidate sourcing is perhaps one of the most challenging and time-consuming recruitment tasks. While social media and job-boards have made sourcing easier, there is still no way recruiters can achieve personalization at scale using traditional tools. AI enables recruiters to automate their sourcing process, reach a wider talent pool, and personalize candidate interactions at scale.
â€œAI is being used to match specialty recruiters, based on their track record by job type, to jobs that need candidates.Â By accessing specialty recruiters with proven success and candidate relationships, employers can quickly and efficiently source quality candidates,â€ says Ken.
Nimish Sharma, CEO and Co-founder at Workex.AIOpens a new window , says, â€œFinding the right talent has always been a daunting task for recruiters. With the help of AI, recruiters are now able to automate their sourcing process as well extend their reach to a larger audience. The deployment of AI in recruitment along with automation technology not only helps in finding the right kind of profiles for a job role but also helps them engage with these profiles using conversational platforms.â€
- Lead nurturing: The next step after you’ve established contact with candidates is to engage and nurture them. Candidate nurturing is fast becoming one of the most critical parts of the recruitment process. Building a talent pipeline with passive candidates brings down recruitment costs and greatly reduces time-to-fill. Here’s where AI can help recruiters create and automate lead nurturing campaigns to deliver hyper-personalized messaging and content to cater to individual candidate needs. Ken says, â€œCRM and marketing automation systems that nurture leads are integrating AI and â€˜chatbots’ with great success. These methods and technologies are also now moving into recruiting for effective lead nurturing, enabling recruiters to spend more time with interested leads versus finding them.â€
- Candidate screening: AI-powered candidate screeningOpens a new window solutions are emerging as a key segment in the AI recruiting space. The idea is to make objective, data-driven decisions when evaluating candidates. AI can reduce or even eliminate human bias when assessing candidates. â€œAI has an opportunity to bring objectivity to talent by focusing its lens on organizations before candidates. Corporations need a mirror to see itself, and AI could be leveraged to filter through the processes and datastreams generated within an organization to reveal problem spots and opportunities early enough for people to take action and create change,â€ says Sarah Nahm, CEO at recruiting software provider, LeverOpens a new window .
- Interviewing: Automated video interviews are probably the best current example of AI in recruiting. This practice goes way beyond standardizing the interview process and saving time. â€œWith automated interviews today, you’ve got much deeper applications such as automated analysis of communications skills and even behavioral analysis of video interviews to predict job performance,â€ opines Ken.
- Onboarding: Personal AI Assistants, or â€œonboarding botsâ€ can now integrate with HR management systems (HCM and HRIS) and essentially act as a new employee’s personal guide to their new workplace.
Given the massive impact that AI can have on recruiting productivity, speed, and cost savings, organizations that adopt AI in their recruiting processes now will gain a major advantage over their competitors. According to Deloitte’s 2018 â€˜State of AI in the Enterprise’ surveyOpens a new window , 42 percent of leaders believe AI will be of critical strategic importance in the next two years.
â€œScout has found that employers using machine learning algorithms in a marketplace recruiting setting have consistently lowered their time to fill by over 30 percent, increased their pool of qualified candidates and fill rates by up to 40 percent, and reduced cost per hire by 30 percent,â€ says Ken.
Key Concerns Around AI in Recruiting: Bias, Diversity, Inclusion, and Privacy
While AI recruiting solutions have certainly matured over the past two years, it hasn’t been smooth sailing all along. One of the major concerns with AI in recruiting has been its tendency to propagate human bias. As a data-driven solution, AI is only as good as the data it has been trained on. When training on historical data, it is likely that AI solutions can pick up patterns that exhibit human bias. If Amazon’s recruiting engine debacleOpens a new window has taught us anything, it is to identify, measure and adjust for human bias in all AI solutions.
â€œWhen AI focuses on identifying talent and matching it to roles, it tends to propagate any bias that exists in the dataset it has been trained on. This means, AI runs the risk of reinforcing the unconscious bias that has existed in recruitment, and possibly even worsening the situation by distancing human judgment even further from the decision,â€ says Sarah. â€œAmazon was right to sunset their AI recruiting engine.Â This is a great example of how AI is only as enlightened as the underlying data it is trained on.Â I don’t believe the technology or data sets are ready for us to try to judge people and flatten them into a score. I do feel optimistic that AI can tell an organization a lot more about its own biases, exactly as Amazon was able to learn in this specific case.â€
Interestingly, AI recruitment solution providers acknowledge that AI development is a tricky job, and it certainly needs expert oversight on issues like diversity and inclusion, algorithmic bias, and data security and privacy.
AI-talent assessment platform, HireVue, recently announced the creation of an Expert Advisory BoardOpens a new window to help guide ethical AI development. â€œAny time AI technology is being developed for use in a context that can have an impact on people’s daily lives, it’s crucial to be approaching its development with ethical principles in mind from day one. As technology evolves over time, ethical principles should be reviewed and potentially updated to ensure that they continue to be meaningful guideposts,â€ says Loren Larsen, CTO at HireVue.
He adds, â€œThe main concern with AI in recruiting is that, unless a process is built to prevent this from the beginning, it’s possible for algorithms to replicate, scale, and even institutionalize human bias. However, with rigorous testing processes that our IO psychologists brought with them from the traditional field of pre-hire assessments, HireVue is able to test for adverse impact against protected groups, remove any data points contributing to such bias, retrain the algorithm, re-test, and repeat until bias is mitigated. It’s possible to remove any problematic data points without losing predictive power only because HireVue Assessment algorithms can incorporate up to 25,000 meaningful data points related to job success.â€
The real opportunity that AI presents in recruitment is scalability and automation to a practice such as industrial/organizational psychology that once relied on expensive consultation services. The predictability and objectivity of talent data can now be democratized and available to all, not just Fortune 500 companies.
Selecting Best-fit AI-powered Recruiting Solutions
As we’ve explored in the section above, training data integrity and quality are the biggest concerns around AI recruiting solutions. â€œWhen evaluating the claims of AI software companies, HR leaders should ask what the training data set is made up from. A tremendous amount of high-quality data is needed for an AI program to be able to provide insights. If an AI program is using resumes and job descriptions, for example, then it’s a pretty poor-quality data set since there’s no inherent validation of that information and most people copy-paste from other sources,â€ advises Sarah Nahm of Lever.
Loren Larsen from HireVue, believes that gauging vendor expertise in AI-powered recruiting is also essential. He suggests five crucial questions to ask AI-recruiting vendors before investing in a solution.
5 important questions to ask about your AI-powered recruiting solution or vendor:
- Does the vendor have deep expertise in the hiring space? Data science capability alone is not enough. AI for hiring relies on a complex set of factors relating to data security and privacy, validation, adverse impact, compliance, and the regulatory environment, accessibility as well as objectivity, and diversity efforts that must be understood in depth in order for a solution to be effective.
- Was the product developed solely by hiring and data science experts, or were social scientists involved? Having IO psychologists as part of the development process ensures that proper testing for impact, efficacy, and candidate experience will be central considerations.
- Does the tool or solution have a track record of success with organizations of your size? Better to be a partner to an AI vendor, not a beta tester.
- What kind of data is being used to train AI models? All training data is not created equal. Some is biased and some is simply not proven by research to be truly predictive of job success.
- Does the vendor audit its algorithms for adverse impact (bias)? One of the biggest promises of AI is its ability to make hiring more objective and fairer. That said, AI is like any powerful technology: improperly built and tested, there is the potential for harm. Vendors should be able to provide full documentation around their process for mitigating bias.â€
In addition to establishing the efficacy of AI recruiting solutions and ascertaining data integrity, HR leaders must also consider the
business outcomes of deploying a smart recruiting solution. â€œThe most important questions should relate to the impact on your goals and return on investment. For example, if your goal is to improve candidate flow, you’ll want to take a look at the recruiting pipeline, see where the bottlenecks are and determine where you need improvement the most. Then, you can look at the expected improvement of an AI solution against the cost.Â Perhaps AI can double your pipeline, but what are the other options and what is the cost of the solution, including implementation, maintenance, etc.?â€ says Ken Lazarus of Scout Exchange.
Recruitment as we know it, is certainly changing and we’ll see new skill-sets and job roles emerge over the next few years. As the adoption of AI recruiting solutions increases, the accuracy and efficacy of these platforms will continue to improve. As with all other business technologies, there is no one-size-fits-all approach to choosing the best-fit recruiting solution. Most vendors offer unique AI solutions for every business segment and we recommend creating a data checklist before investing in an AI recruiting solution. Lastly, measuring the success of AI is no different than any other business initiative. â€œIt gets back to understanding your objectives and setting measurable goals from the outset. For Scout, that means meeting our own goals of greater numbers of high-quality, satisfied placements happening faster than ever, as well as meeting clients’ goals â€“ typically related to reach, speed, cost, and diversity,â€ says Ken.
Words to set an effective strategy by!
In the next episode of our â€œSucceeding with AI in HRâ€ series, we’ll dive deeper into how AI is transforming Human Capital Management, how you can deploy AI-powered HCM solutions and measure the ROI of your AI initiatives.
Read our previous issue: The Beginner’s Guide to AI in HROpens a new window
What trends are you tracking in the AI recruiting space? Let us know on Twitter or LinkedIn or Facebook. We’re always listening!