Yixuan Xie
Medill Reports

Hiring the right people is crucial to an organization’s success and companies are turning to artificial intelligence to optimize that process.

AI, which uses computer science to make machines imitate human intelligence and behavior, is revolutionizing numerous industries. It is the technology behind Amazon Inc.’s cashier-less stores, Tesla Inc.’s self-driving vehicles and Apple Inc.’s Siri voice assistant. It is also lending a helping hand to the recruiting industry to find the right people for the right jobs.

The U.S. Department of Labor reported an unemployment rate of 3.7 percent for July, which is near a five-decade low. Despite being good news for job seekers, the rate has some employers desperate to find talented workers when so many people are already employed.

There are typically two ways to fill vacant positions. One is posting a job and waiting for people to apply. The other is turning over every rock to find candidates with the right skill sets and engage them, which is known as sourcing, said Ben Eubanks, a human resource analyst and author of “Artificial Intelligence for HR: Use AI to Support and Develop a Successful Workforce,” who cited the latter as more common even though it is time and labor intensive. Sourcing for a highly technical role can take anywhere from eight to more than 30 hours, Eubanks said.

There are already different solutions in the market which identify search criteria and find candidates that match the capabilities employers are looking for in big databases. While that is an improvement compared to manually looking for keywords on resumes, Eubanks said it’s the introduction of AI that has changed recruiting like never before.

Augmenting Reach

Far beyond keyword matches, AI expands a search by recommending related skills to help recruiters find candidates they would not have otherwise found, Eubanks said.

Sunny Son, executive recruiter at Humana Inc., a health insurance company, has been using an AI-powered sourcing tool developed by Mountain View, California-based startup Hiretual Inc. since 2017.

When entering a keyword during the search, Hiretual’s AI technology identifies relevant skills and searches for additional layers of skills. This allows recruiters to seek candidates they would not have known how to search for a number of reasons, including not being knowledgeable about programming languages.

“For example, if I put in ‘Python’ during my search, Hiretual would ask me if I want to add PuDB, Java or SQL, which are relevant skills to that,” Son said. “So, it’s thinking for me.”

Delivering Qualified Candidates

 AI can also learn from user interactions by automatically applying to the next search their feedback on whether or not the candidates shown in the first search are good fit, said Ninh Tran, founding member at Hiretual.

“Based on your input on why they are not a fit, we are able to search through candidates and deliver very highly targeted results,” Tran said.

Unlike job boards which show recruiters everyone who applies, Tran said Hiretual only delivers 0.15 percent of the candidates that best match the requirements.

By reducing the manual work of sorting and filtering, Son said a search which used to take her 10 hours to refine the results only needs two hours to get the same result with Hiretual’s AI technology.

“That’s tremendous time saving,” Son said. “And you also have more qualified and diverse results because Hiretual is pulling from all types of sources.”

Hiretual searches candidates from more than 30 public platforms like LinkedIn Corp., Facebook Inc., Github Inc., Twitter Inc. and connects different pieces of data to individuals to understand their professional experience, expertise, availability and market value, Tran said.

Trish Wyderka, research associate at Fidelis Companies LLC, another Hiretual customer, echoed Son’s comments. She said she always has to do multiple candidate searches for different industries at the same time.

“Hiretual is already multitasking enough,” Wyderka said. “So, I feel like it has cut my work by more than 50 percent.”

Improving Response Rate

LinkedIn has also started leveraging AI to engage with potential candidates more efficiently, which is equally important as identifying them in today’s competitive market full of recruiters.

LinkedIn used to show every recruiter who does the same search the same list of candidates that best fit the skills they are looking for, Eubanks said. As a result, people on that list are constantly getting contacted by all the recruiters, and they’re not going to respond, Eubanks said.

“When recruiters are sending InMails, which is LinkedIn’s private email system and recruiters pay a lot of money to be able to send InMails, all recruiters know that the response rate is very poor,” Eubanks said. “In their database, the response rate is about 9 percent. That’s very expensive as a medium for conversation.”

To solve that problem, LinkedIn Recruiter, a platform for finding and connecting with candidates, prioritizes people who are most likely to respond, said Neha Mandhani, senior product marketing manager at LinkedIn. Its AI-driven system is not only able to provide better recommendation of candidates based on whether recruiters save or hide a profile, but it also looks at hundreds of signals of what those candidates’ interests are on LinkedIn. For example, if a person turns on the “open to new opportunities” feature or shows interest in a company in different ways, that’s a signal, Mandhani said.

“Whether it’s following the company, following its CEO or engaging with its content on LinkedIn, these are strong predictors of a person being interested in a company and if relevant to the recruiter, AI will push that candidate’s profile higher in a search result,” Mandhani said.

Challenges

Despite several benefits, concerns over privacy have peaked with the rise of AI. Among more than 5,000 consumers across the U.S., the U.K., and Australia, 71 percent said they don’t want companies to use AI that threatens to invade their privacy, even though it improves the customer experience, according to consumer research released in December 2017 by professional services firm Genpact LLC. The study also showed that 59 percent of people believed their governments should do more to protect their privacy from being encroached upon by AI companies.

To predict whether a person who is applying for a job or for whom employers are searching fits a certain profile model, much personal data needed, which is now often mined from social networks and taken from data brokers, said Mireille Hildebrandt, a research professor on “Interfacing law and technology” at the Vrije Universiteit Brussels. Gathering that much information, Hildebrandt said, may be a privacy invasion.

“But I think what is worrying people more is the extent that people become manipulable,” Hildebrandt said.

Companies that develop AI tools are likely to keep the data they use and how they create the models as a trade secret or as intellectual property rights in order to make money, Hildebrandt said. That means neither the recruiting company nor the potential candidates can assess whether they are actually reliable, which Hildebrandt said is a bigger problem.

Because of controversies around data privacy, the European Union (EU) passed the General Data Protection Regulation (GDPR), applicable since May 2018, that requires organizations, regardless of their location, to have an explicit purpose to process or hold any personal data of someone in the E.U., Hildebrandt said.

AI-powered recruiting platforms have been impacted by the GDPR. According to reviews of Hiretual on Capterra, an online peer review site, the loss of value caused by the GDPR compliance was a top compliant from customers. There were comments made by anonymous reviewers such as “this tool became useless after the GDPR” and “if the person is from the E.U., then Hiretual will not be able to get you their info.”

To meet the GDPR compliance, Tran said Hiretual has spent the past year and a half to gather consent from candidates through partnerships with job boards. It has also applied legitimate interests of helping people find jobs as a lawful basis for processing data.

“I would say about 70 percent of candidates on our platform have already given us consent or we have legitimate interests for,” Tran said. “It is going to increase over time.”

Hildebrandt said the GDPR will contribute to eradicating applications that fail to do what they claim and are unable to prove the necessity to legally gather and process personal data.

“Let’s say there are 350 relevant AI applications on the market, and it’s not going to be easy to figure out which are totally crap and which are really good. The GDPR will help,” Hildebrandt said. “The regulators have the competence to go to companies and involve technical people to check. That basically helps the market to become a good functioning market, though it may take some time before companies are fined and become aware of this.”

The U.S. has also started on a path toward privacy regulation. With California Consumer Privacy Act (CCPA) and Nevada’s new data privacy law passed and bills planned in many other states, these AI recruiting solutions may be further impacted as the collection, use and sharing of personal data faces more limitations, which would narrow the sourcing pool and make it much harder to provide candidates’ contact information for recruiters to better engage with promising ones.