This article was originally published by AI Frontier.
dwz.cn/7yzqN7


Planning editor: Natalie


By Stephen Buranyi


Compiled & Edited by Debra

The automation revolution is sweeping the recruitment world, with everything from a job candidate’s facial expression to his voice now being analysed using algorithms and artificial intelligence. But do you understand the cost of sacrificing the diversity of the workforce and the interests of workers? Maybe while you’re still trying to apply for a job, the robot has already rejected your application for the HR and won’t even qualify you for an interview.


What we see now: AI recruitment solves many problems

Indeed, in an era of increasingly high human resource costs, ARTIFICIAL intelligence can solve many problems that human HR cannot solve.

According to a survey by global hr agency Hays, there is a huge gap between the hiring plans of global companies and the skills shortage of their employees in 2018, with 92% saying the problem has significantly affected their production and employee satisfaction and spending.

Moreover, some jobs, such as algorithm engineers and data scientists, have always been in short supply, making it difficult to acquire such scarce and competent people.

Faced with these problems, many companies are turning to AI.

According to Deloitte Human Capital Trends, 38% of enterprises have adopted AI technology, and this proportion will reach 62% in 2018.

Deloitte Bersin reports that companies using AI are more successful in predictive analytics and technology tools than those not using AI, and estimates that the former generate 18% more revenue and 30% more profit than the latter.

In addition to helping with sourcing candidates, AI can reduce bias in the hiring process. What’s more, most companies prefer “passive candidates,” those who are still working and don’t have a strong desire to leave. AI bots can now find such people from personal blogs, groups, chat rooms and other traditional social networks, and they can match jobs much more efficiently than people.

In addition, THE AI algorithm can also track the update of resume description at any time, track the people who have the intention of changing jobs, liberate HR from the heavy work of searching candidates, optimize job requirements and provide accurate description of talent pool, promote the recruitment process faster, improve the speed and quality of recruitment.

In 2017, for example, Google launched an AI job search function that allows job seekers to enter a keyword and get a list of recommendations for job openings. To create this list, Google must first remove duplicate listings that employers post to all of these job sites. Its machine learning training algorithm then filters and classifies them. These job sites often use specific tags to help search engines understand certain job postings.

It is reported that the domestic pull net has also launched a new identity verification mechanism for enterprises and recruiters. On the basis of the original business license, enterprise email, financial account verification, the introduction of Baidu AI collaboration plate, enterprise HR registration not only needs to complete remote identity information collection, but also need to verify real identity through face recognition. Strengthen user recruitment security through “multiple verification” of enterprise and HR qualifications, and crack down on false information and fraud on recruitment platforms.

Zhaopin also launched a new product in 2017, with functions such as “job refresh” and “top” to expose a lot of positions. The “Chat” function allows job seekers to communicate with HR in real time, and the recommendation of relevant positions has reached a relatively accurate level, which has solved many problems for employers and job applicants.

But it also makes it harder to find a job

But is AI really that powerful in the hiring process? The answer is no, at least from the current level of recommendation and screening systems on various platforms, otherwise there would be no “underground anti-automation war”.

According to psychologist Nathan Mondragon, experiments have shown that there are thousands of small details we can use to find the right person for a job. Mondragon is the lead psychologist at Hirevue, which sells software that uses algorithms and artificial intelligence (AI) to screen job applicants.

Hirevue’s flagship product is used by global giants such as Unilever and Goldman Sachs, which require candidates to answer standardised interview questions in front of a camera. At the same time, the software, like a team of hawk-eye psychologists hidden behind a mirror, takes in thousands of notes about a candidate’s posture, facial expression, tone of voice and word choice.

“We asked people to break down candidates’ answers into thousands of data points for verbal and nonverbal cues,” Mondragon said. “If you’re answering a question about how you plan a $1 million budget, your eyes will tend to move up and go silent at the same time, or make ‘um’ and ‘ah’ sounds. Your head and eyes will tilt up slightly at the same time. Facial movement analysis tells us that you are entering a state of creative thinking.”

The program converts this data into scores, which are then compared with the scores of well-performing employees that the program has “learned.” The idea is that a potential top performer will have similarities to someone who has now proven to be a top performer, traits that a human interviewer might not notice.

As far-fetched as it sounds, methods such as speech analysis and reading “microexpressions” have been used in policing and intelligence before, without notable success. Mondragon says its automated analysis matched the results of personality and ability tests, and that clients did report improved employee performance and reduced spending.

“It’s kind of dehumanizing, and you may never get access to an employer.”

— Robert, ordinary worker

Hirevue is just one of a number of new companies applying ARTIFICIAL intelligence to recruitment, hoping to reduce the cost of hiring human resources. Hirevue estimates that the “pre-hire appraisal” market is worth £2.14 billion ($3 billion) a year. According to Doug Rode, senior managing director of recruiter Michael Page, the number of service companies selling the AI family has increased significantly in the past year, but the quality is uneven.

A study last year by the Chartered Institute of Personnel and Development found there were an average of 24 applicants for a low-wage job. The UK’s largest private employer received more than three million job applications in 2016. As the number of people applying for jobs has increased, Tesco has tried to reduce human resources as much as possible and has instead used more and more decision-making automation. For more than a decade, they began using a simple program that scans resume text for keywords such as a candidate’s education, skills and past employers to flag recruiters. The process has since expanded to include confusing tests, psychometric tests and custom games to reject applicants before meeting employers.

This has revolutionized the way many people interact with prospective employers. Where once a job seeker could approach multiple companies with a standardised resume, the burden has shifted from employers to job seekers, and human interaction has played less of a role in hiring, making an already difficult process worse.

In addition to the confusing and dehumanizing experiences that automation and AI can lead to (potentially unequal data), there is also the risk of “discrimination” by algorithms, such as job seekers who send out thousands of resumes and then receive no response. It could be that your resume is buried in a pile of resumes that HR can’t see, but it could also be that robots have taken over and sifted through your resume like garbage, never seeing candidates again. What should you do? To what extent do these algorithms have bias? How do software developers and licensed third party companies handle your personal data? Will the circumstances be so special that less competitive candidates, or those less skilled in new technologies, will be eliminated without even qualifying for an interview?

What the average job seeker thinks about AI recruitment

“These are artificial barriers that make the recruitment process seem insurmountable,” says Heather Davies, a retired human resources coordinator and co-organiser of the Christian Working Against Poverty Club. While it’s inevitable that job seekers accept that automation will become more common — it’s 2018, after all — it’s the absence of human communication that’s really frustrating.

“It’s dehumanizing, and the candidate may never get to the employer,” says Robert, an average worker in his forties who finds temporary work through work committees and recruiters. Harry, 24, has been looking for a job in retail for four months. In retail, almost every job application requires some kind of test or game, from personality tests to math tests, to screen out applicants. He completes tests four or five times a week. Usually the rejection comes immediately after the test, but sometimes the email is deliberately delayed to make it look like they’ve taken the time to decide if you’re right for the job. But this is all an illusion.

“It’s frustrating. You never know what you did wrong to make you feel a little sad. “Said Harry.

“It’s a big hurdle. Why suddenly do you require an older bricklayer to have IT skills?”



— Lynda Pennington, Croydon Post Club

Older job seekers face more complicated problems. Many rely on council or volunteer support to help them fill out application forms or submit resume forms.

Kirsty McHugh, director of the Association for Employer-Related Services (ERSA), expressed concern that apps “exclude non-traditional candidates without a second thought,” adding that ERSA members would not encourage employers to use them.

Most of Britain’s unemployment figures ignore the large number of people who have given up looking for work. According to ONS figures, 8.7m people aged 16 to 64 are “economically inactive”, meaning not in work or looking for work. We can’t predict how many people will be put off looking for work by the new hiring methods.

Even among the largest groups where automated recruiting is seen as successful, there are problems. Several professional recruiters told me that, for every job, many candidates are put off by these systems. While 58 percent of North American job seekers said they would be happy to interact with automation projects, according to a survey by recruiting agency Allegis Global Solutions, this ambiguous statistic was widely interpreted as a green light for political correctness.

Deborah Caldeira, a master’s student at the London School of Economics, says that after 86 failed job applications in the past two years, including Hirevue, she has become disillusioned with automated systems. With no one sitting at the table and “no real conversation or communication,” it’s hard to know “what kind of employee the robot is looking for,” she said.

Despite her excellent grades and extracurricular activities, she became unsure of herself and doubted her abilities.

As automation and AI filter is more and more far, some applicants fought the war against automated underground, they exchange in the network BBS employers test answers, and create a false application to limit their processes, even in the heart of the application “Oxford” or “Cambridge” and other words in the resume (white) hidden in the text, in order to filter through automation.

It is difficult to determine how widely automation is used in recruitment, mainly because companies are reluctant to disclose their use, including Sainsbury’s and Tesco, some of the UK’s biggest employers.

“People are becoming more aware of these systems and may protest AI decisions.”



— Sandra Wachter, Oxford Internet Institute

Strengthening supervision and AI+ human resources is the right way to explore talents

While incorporating AI into the hiring process is certainly a good thing for both employers and job seekers in the long run, there are real issues that need to be addressed for job seekers.

According to Christina Colclough, head of digitisation and trade at UNI Global Alliance, legislation is actively promoting a balance between the use of AI in the recruitment process and applicants, although unions have been slow to respond to technological change, But UNI Global and other companies are developing a variety of workers’ digital rights charters that govern automation and AI-based decision-making and incorporate them into group agreements.

A forthcoming update to the EU’s General Data Protection Regulation (GDPR) will require companies to disclose information about every AI decision automation technology that “significantly affects individuals”, with applicants also having the right to challenge decisions or request human intervention.

However, GDPR has limited binding power. For example, if a company’s human resources were involved in the decision making process, even if only a little, the applicant does not have the right to continue to challenge the company’s decision, and they do not have the right to ask the company to explain the technical details of the decision.

“This is very difficult legislatively because these programs are highly technical, very complex and difficult to understand, even for the experts who build them, and how they work is often protected by Copyrights held by companies.” Christina says.

Importantly, the GDPR has highlighted bringing the recruitment industry into the regulatory fold, which will give candidates more say in the recruitment process, people will be better informed about these systems and have the right to challenge the results of AI decisions.

Finally, combine the AI and manpower recruitment method, is the right path for recruitment problems, finally to decide whether to hire an interviewer shouldn’t be a machine, but human HR with rich experience and experience, and the machine can play the role of guidance and the transformation in the process of recruiting, how businesses and the interviewer can save money and time costs.

Original link:

www.theguardian.com/inequality/…

www.forbes.com/sites/group…

Techcrunch.com/2017/06/20/…

For more content, you can follow AI Front, ID: AI-front, reply “AI”, “TF”, “big Data” to get AI Front series PDF mini-book and skill Map.