Amazon canceled a multi-year project to develop an experimental automated recruiting engine after the e-commerce giant’s machine learning team discovered that the system was exhibiting explicit bias against women, Reuters reports. The engine, which the team began building in 2014, used artificial intelligence to filter résumés and score candidates on a scale from one to five stars. Within a year of starting the project, however, it became clear that the algorithm was discriminating against female candidates when reviewing them for technical roles.
Because the AI was taught to evaluate candidates based on patterns it found in ten years of résumés submitted to Amazon, most of which came from men, the system “taught itself that male candidates were preferable,” according to Reuters:
It penalized resumes that included the word “women’s,” as in “women’s chess club captain.” And it downgraded graduates of two all-women’s colleges, according to people familiar with the matter. They did not specify the names of the schools. Amazon edited the programs to make them neutral to these particular terms. But that was no guarantee that the machines would not devise other ways of sorting candidates that could prove discriminatory, the people said.
The company scuttled the project by the start of 2017 after executives lost faith in it. By that time, however, it may have already helped perpetuate gender bias in Amazon’s own hiring practices. The company told Reuters its recruiters never used the engine to evaluate candidates, but did not dispute claims from people familiar with the project that they had had looked at the recommendations it generated.
In the latest release of its applicant tracking system, Google Hire, the tech giant has added three new features that use Google AI to reduce repetitive and time-consuming tasks, Senior Project Manager Berit Hoffmann wrote in a blog post announcing the update on Tuesday. From measuring user activity, Hoffmann noted, Google determined that Hire has already cut down the amount of time recruiters spend on common tasks like reviewing applications or scheduling interviews by up to 84 percent; the new features are intended to simplify the process even further. The new features include:
- Interview scheduling: When a user wants to schedule an interview with a candidate, Hire now uses AI to automatically suggest interviewers and time slots. The AI will also alert the recruiter if an interviewer cancels at the last minute and recommend a replacement. “This means hiring teams can invest time in preparing for interviews and building relationships with candidates instead of scheduling rooms and checking calendars,” Hoffmann writes.
- Résumé highlighting: To reduce the amount of time recruiters spend scanning candidates’ résumés for key terms, Hire will now automatically analyze the terms in a job description or search query and automatically highlight them on résumés. Google introduced this feature after observing that users were frequently using “Ctrl+F” to search for the right skills—an easily automated process.
- One-click candidate phone calls: The final feature is designed to make it easier for recruiters to reach out to candidates by phone with a click-to-call functionality. Users can call a candidate simply by clicking on their phone number, while the system will automatically log calls to keep track of which candidates have already been contacted by whom.
Google Hire was launched last July as part of the company’s G Suite of enterprise software offerings, but only for US businesses with under 1,000 employees: a deliberate decision to help level the playing field for small and mid-sized businesses that lack the dedicated recruiting resources and bespoke applicant tracking systems of their larger peers. In April, Google Hire expanded into the sourcing realm with a new “candidate discovery” feature that allows users to more easily keep track of past candidates who might be good fits for newly open positions, along with more advanced search capability to provide more relevant results based on what recruiters are actually looking for.
ERE’s Joel Cheesman sees these new AI enhancements as further evidence of the massive edge large tech companies like Google, Facebook, and Microsoft enjoy in the new era of online recruiting: “Deep integration into technologies that so many people already use daily, such as Gmail and Google Calendar, must drive traditional recruiting technology solutions crazy. Build all the Chrome extensions you want, but nothing’s ever going to be better than the stuff Google has baked itself.”
In the latest of this year’s big waves in recruiting technology, the Japanese HR conglomerate Recruit Holdings has finalized a deal to acquire the recruiting, job review, and salary transparency site Glassdoor for $1.2 billion, GeekWire’s Taylor Soper reported on Tuesday night:
Glassdoor, founded in 2008, will remain a “distinct and separate part” of Recruit Holdings’ HR technology business segment. The Tokyo-based company has more than 45,000 employees; its last big acquisition was swooping up jobs site Indeed in 2012. The all-cash deal is subject to regulatory approval, expected this summer.
CEO Robert Hohman will continue to lead the company. The acquisition is in line with Glassdoor’s longstanding vision of becoming a world-leading recruiting platform, Soper notes, pointing to remarks co-founder Rich Barton made at a Zillow event in 2014:
“Our BHAG (Big Hairy Audacious Goal) for Glassdoor is to become the largest recruiting company in the world, to help everyone find a job and company they love, to become ‘TripAdvisor for employment,’” he said in 2014. ” … This is a revolution in the jobs industry. One day we will become the most important company, the most important marketplace, in recruiting.”
Recruit being the owner of Indeed (as well as SimplyHired, another major job search site), it is natural to speculate that it might combine these massive properties into an even larger online recruiting behemoth. Hisayuki Idekoba, Recruit’s chief operating officer, says there are no plans to integrate Glassdoor and Indeed, but they may partner on “specific challenges,” Bloomberg’s Alex Barinka adds.
No one ever intends to create a biased algorithm and there are huge downsides for using one, so why do these algorithms keep appearing, and whose fault is it when they do? The simplest explanation for why algorithmic bias keeps happening is that it is legitimately hard to avoid. As for the second question, there is no consensus between algorithm developers and their customers about who is ultimately responsible for quality. In reality, they are both to blame.
Vendors and in-house data science teams have a lot of options for mitigating bias in their algorithms, from reducing cognitive biases, to including more female programmers, to checklists of quality tests to run, to launching AI ethics boards. Unfortunately, they are seldom motivated to take these steps proactively because doing so lengthens their timeframes and raises the risk of an adverse finding that can derail a project indefinitely.
At the same time, clients are not asking for more extensive oversight or testing beyond what the developer offers them. The client usually doesn’t know enough about how these algorithms work to ask probing questions that might expose problems. As a result, the vendor doesn’t test or take precautions beyond their own minimum standards, which can vary widely.
In a recent interview with Employee Benefit News, HireVue’s Chief IO Psychologist Nathan Mondragon discussed a situation in which his company built a client an employee selection algorithm that failed adverse impact tests. The bias, Mondragon said, was not created by HireVue’s algorithm, but rather already existed in the company’s historical hiring data, skewing the algorithm’s results. In his description, they told the customer: “There’s no bias in the algorithm, but you have a bias in your hiring decisions, so you need to fix that or … the system will just perpetuate itself.”
In this case, Mondragon is right that responsibility for the bias identified in the adverse impact test began with the client. However, I would argue that vendors who do this work repeatedly for many clients should anticipate this outcome and accept some responsibility for not detecting the bias at the start of the project or mitigating it in the course of algorithm development. Finding out that bias exists in the historical data only at the adverse impact testing phase, typically one of the last steps, is the developer’s fault.
Google’s powerful new job search feature, launched in the US last June, has begun its global expansion and is now available in India and Canada. The India expansion will aggregate job listings from over a dozen partners, the Economic Times‘ Surabhi Agarwal reported last week, including some multinational partners like LinkedIn and IBM Talent Management Solutions, as well as India-specific job search sites like QuezX, QuikrJobs, and Shine.com:
Rajan Anandan, Vice President India & Southeast Asia, said in the last quarter of 2017, Google saw more than a 45% increase in the number of job search queries. “SMEs are the largest job creators but are often unable to make their listings discoverable. This new job search experience powered by our partners and our open platform approach attempts to bridge this gap,” he added.
With “Google for Jobs,” as the feature is commonly known, the search giant does not host job listings itself but rather directs search traffic to partner job boards using its sophisticated search algorithm, promising to more efficiently connect job seekers with positions already being listed in their geographic area and professional field.
Canadians can also now use Google’s powerful search tool to find their next job, the company has announced. Partner organizations in that country include the Canadian government’s Job Bank/Guichet-Emplois, BCJobs.ca, LinkedIn, Glassdoor, Monster.ca, Jobillico, and Jobboom.
You’ve heard of 360 reviews, but what about 360 previews? At SHRM last week, Lin Grensing-Pophal took note of the novel recruiting technology, through which recruiters can give prospects a realistic virtual view inside their potential future workplace. Candidates even get a chance to see exactly what the jobs they are applying for entail:
The content of 360-degree videos can vary: They can offer a “day in the life” perspective about a specific job, interviews with employees and others from the organization, or a bird’s-eye view of company activities such as events and town hall meetings. The management training program at Compass Group North America, a family of food-service and support-services companies serving the hospitality industry, for instance, allows viewers to explore the company’s facilities through a tour narrated by those who are in the program. Viewers see the facilities and learn about the incumbent’s experience in the role. The viewer’s vantage point can be swiveled around for a 360-degree experience.
Diversity, new interviewing tools, data, and artificial intelligence are the four trends set to have the biggest impact on recruiting in the coming year, according to LinkedIn’s latest Global Recruiting Trends report. Based on a survey of over 9,000 talent leaders and hiring managers worldwide, along with a series of expert interviews, the report underscores the growing role of technology in shaping how companies meet their hiring goals, of which diversity is increasingly paramount. Nonetheless, while many HR leaders see these trends as important, the number of organizations fully acting on them lags far behind.
Diversity was the top trend by far, with 78 percent of respondents saying it was very or extremely important, though only 53 percent said their organizations had mostly or completely adopted diversity-oriented recruiting. In recent years, diversity has evolved from a compliance issue to a major driver of culture and performance, as more and more organizations recognize its bottom-line value. This shift was reflected in the LinkedIn report, with 62 percent of the companies surveyed saying they believed boosting diversity would have a positive impact on financial performance and 78 percent saying they were pursuing it to improve their culture. Additionally, 49 percent are looking to ensure that their workforce better reflects the diversity of their customer base.
Diversity was the only top trend identified in LinkedIn’s survey that wasn’t directly related to technology, but technology is definitely influencing how organizations are pursuing it. In the past year, we have seen the emergence of new software and tools to support diversity and inclusion. The aim of these tools is to remove the human error of unconscious bias from the recruiting process, but it’s important to be aware that automated processes can also develop built-in biases and end up replicating the very problem they are meant to solve. This is an issue we’ve been following in our research at CEB, now Gartner; CEB Diversity and Inclusion Leadership Council members can read more of our insights on algorithmic bias here.
The development of new interview tools and techniques was identified as the second most important trend, with 56 percent saying it was important. The LinkedIn survey found that the most common areas where traditional interviews fail are assessing candidates’ soft skills (63 percent), understanding candidates’ weaknesses (57 percent), the biases of interviewers (42 percent), and the process taking too long (36 percent). The report highlights five new interviewing techniques, all enabled by technology, that aim to address these problems: