The human recruiting process has a long timeframe and high complexity, but can simplified into four key words: searching, screening, interviewing and hiring. Today’s large global labour force and convenient online application systems however have put a stain on the screening part of the recruiting process. An average of 250 resumes are received for each corporate job opening — a number that continues increasing — and it takes an experienced recruiter about five minutes to review each resume. Moreover, in spite of the time-consuming resume reviewing process, hiring remains plagued by randomness and uncertainty often due to unconscious human subjectivity and biases.
Many companies around the world are asking the same question: How to objectively and efficiently identify the best candidates from a huge pile of resumes?
Artificial intelligence can help companies navigate a global human resource market that is projected to reach US$30 billion by 2025, due in large part to technological proliferation. A trained AI system is capable of scanning resumes, extracting keywords, scoring and ranking candidates, and even helping with interviews.
Founded in 2013, Ideal is a Canadian startup that offers machine learning based human recruiting solutions ntegrating its virtual assistant with a client’s applicant tracking system (ATS). The virtual assistant was trained on “millions of past hiring decisions” and can screen resumes by learning the client’s current employees’ experience levels, skills and other qualities, then applying this understanding to rank job applicants, producing better predictions than a simple ATS.
“Intelligence-driven talent acquisition” is the selling point of ARYA, the flagship tool of US recruiting robotics software company LeoForce. Instead of passively screening resumes that have been sent to a client, ARYA is tasked with creating connections between the company and the talent it might need. Using machine learning, ARYA screens millions of online profiles to build a talent pipeline and unified recruiting ecosystem. ARYA maintains a talent pool of some 40 million candidates and 6,000 global recruiters, and provides services to over 10,000 client companies worldwide.
Because screening and interviewing are interconnected processes that must be jointly addressed, researchers are also providing AI models with “interview” training. Like Ideal’s virtual assistant, the Avrio AIis able to scan and rank resumes. But a distinctive Avrio feature is its Facebook chatbot, which uses natural language processing technology to “talk” with candidates in a sort of pre-screening interview. The chatbot asks and answers questions, and analyzes candidates’ word choices and speech patterns to predict whether they would be a good fit for the client.
Facial and speech recognition technologies enable San Francisco startup VCV.AI to provide recruitment services that enter the realm of science fiction. Their recruiting robot can conduct both phone and video interviews with candidates. By analyzing speech patterns, facial expressions and gestures using a deep learning neural network, the robot provides more in-depth analysis than a simple chatbot. And that’s not all, with every interview it does the model also checks and improves its algorithm and learning patterns.
Slowly but surely, AI is transforming the human recruiting process. So how can candidates in the application pool deal with this new reality?
It’s important to understand the differences between AI and human recruiters. AI is superior to human recruiters in automating large-volume tasks and improving hire quality via standardized job matching. There are however also drawbacks to AI recruiting systems. A huge amount of data is required to build a mature AI system, and this can be quite costly and time-consuming. Also, AI can reduce but not completely eliminate human bias. For instance, if a company has more male than female employees, a supervised machine learning AI system with no regularization term can easily favour male candidates to match the current company identity. Also, because resumes contain personal information, a poorly secured AI system could result in serious privacy leakages.
Today’s job applicants can do a few things to favourably position themselves in the AI in HR environment. Ensuring your resume can be converted to a computer recognizable variable is the most basic and important rule. Sometimes, in order to create more attractive page formatting, a candidate may convert some of their resume content into an image, which may not transmit the content to the AI effectively. If important information is not accurately detected by the AI system that may result in a lower ATS or other ranking. Also, because the AI system reading your resume may be using a natural language processing algorithm, it is a good idea to make your resume as clearly descriptive as possible. Finally, if you feel the rejection of your resume may be due to bias in the AI screening system, do not hesitate to communicate with the company and bring up your concerns. Companies are acutely aware of the potential for problems in these new systems and your feedback provides them an opportunity to improve their system. They are also likely to have a human take a second look at your resume or interview you in person, which provides you with an additional opportunity to get the position.
Author: Linyang Yu | Editor: Michael Sarazen