Tag Archive: AI

  1. Looking Candidates in the Eye: Compliance in Video Interviewing

    Before video interviewing software, everything from the initial phone screen to the live interview was a low-tech operation. Now, thanks to interview technology, we have efficiency, and we have options. Options for candidates to interview during off hours from the comfort of their own space. Options for recruiters to review candidates throughout the day, rather than in predetermined time blocks. Options for hiring teams to collaborate and share feedback in real time.

    Of course, with these options come questions: Do video interviews promote bias? What does compliance look like? How do employers ensure it? And what does all this mean for candidates? We’ve seen these questions come up repeatedly over the years, especially as the technology continues to advance, incorporating new features around AI and facial recognition. So let’s get to the bottom of bias and compliance.

    The truth about the B-word

    Bias is something organizations are taking big steps to avoid, and for a good reason. Besides being bad for your reputation, more than one study points to the benefits of a diverse workplace, including one that shows a 20% increase in innovation. Still, when it comes to interviewing (or anything), change makes people uneasy – especially when we’re talking about inserting a piece of technology that will influence a hiring outcome.

    The question of how to avoid bias in video interviewing first came up about ten years ago, and the Equal Employment Opportunity Commission (EEOC) responded by offering the following guidance: “Before using video resumes and other video screening devices, a covered entity should proactively formulate and communicate to selection officials how the video resumes can assess specific qualifications and skills that are necessary for success in the position. Additionally, a covered entity could require that several people assess each video resume in relation to the stated job requirements.”

    In the years since, video interviewing has come to provide much-needed structure and alignment for everyone involved in hiring. So much so that today’s solutions are likely less biased and more defensible than traditional, unstructured interview methods.

    Keeping up with compliance

    Like bias, compliance is another area of concern, and somewhat of a moving target for employers. What does it mean to be compliant? The answer to this question keeps evolving, in light of recent legislation. Following the 2010 EEOC letter above, the commission revisited video interviewing several times, most recently in 2018. Here, the EEOC explored video (or “digital”) interviews in the context of the Americans with Disabilities Act (ADA). Again, the EEOC reiterated that interview technology, specifically digital, does not violate any existing legislation, going on to recommend that employers include language inviting candidates to contact them (the employer) with any concerns.

    With the enforcement of GDPR and newly passed Illinois Artificial Intelligence Video Interview Act, compliance goes even further and now includes the candidate’s explicit consent. Unlike in-person meetings or even phone screens where consent is implied in the invitation, under these laws, scheduling a video interview requires a candidate’s permission before moving ahead. That’s first and foremost.

    Then, there’s the even larger topic of AI and the underlying logic that’s being used for selection decisions. With hiring algorithms and facial recognition software being built into video interviewing platforms, legislators have called for greater transparency, prompting solution providers and employers to re-think any black box methods they have in place.

    Best practices going forward

    We know from years of video interviewing that structure reinforces the process (and the results!) This differs immensely from phone screens and face to face interviews, where hiring teams are unlikely to ask the same questions of every candidate, letting impromptu conversation guide the way instead. With video interviews, each candidate gets the same questions, the same time to think, and the same time to answer. When reviewing responses, your panel uses an evaluation form to identify core competencies, values, and so on. This approach reduces bias, and ultimately liability, when compared to unstructured methods.

    That said, newer technologies in video interviewing, particularly those leveraging facial recognition software, demand continued conversation and consideration. That’s why states like California are considering limiting use until vendors can guarantee the purpose and efficacy of these features. While we wait to see where legislation and innovation take us, it’s in the best interest of hiring organizations to arm themselves with knowledge and resources, which can then be passed on to candidates. Following EEOC guidelines might mean you err on the side of over communicating the purpose and intended use of any technology used for hiring. This will help eliminate any lingering doubts and ensure compliance all around.

    For a deeper dive into bias, compliance, and best practices when using AI, watch: AI, Algorithms, and How to Make Great HR Tech Investments.

  2. AI – What’s Next for Video Interviewing Software?

    Video interviewing software has taken Talent Acquisition by storm, and the market is buzzing with excitement about future capabilities.

    Early adopters of specialized recruiting tech found a secret weapon in video interviewing software. Finally! Interviews could be done in a structured and scalable way, in less time, across a broader and more diverse candidate pool.

    Today, more than 60% of companies use video interviewing, according to an OfficeTeam survey, and adoption is on the rise. Hiring teams are jumping at the chance to connect with more candidates, and put the more time-consuming, monotonous screening methods behind them.

    Jose Alcantara, HR Manager at MSX International, says thanks to video interviewing,

    “We’ve found the solution to our high-volume recruitment challenges!”

    Rosie Alonso, Director of Talent Acquisition at Tech Data, echoes this excitement, saying,

    “I could not imagine recruiting in today’s environment without it!”

    While video chat and video conferencing make it easy to meet virtually, even more powerful is the ability to do pre-recorded video interviews. Hiring teams can review at least 3 pre-recorded interviews in the time it would take to conduct one 30-minute phone screen.

    That’s an efficiency boost of 3X! But, employers know that today’s technology can do more than boost efficiency.

    Enter AI

    Technology that can accelerate outcomes is good. Technology that can anticipate and influence outcomes is even better. That’s what makes AI such a powerful tool.

    So what’s the future for AI and video interviewing?

    According to a Harvard Business Review article, AI algorithms are being used to mine data – including tone of voice, gestures, and facial expressions – from video interviews to make predictions about a candidate’s job potential.

    Humans are constantly interpreting body language and social cues, and in just a few seconds, we can learn a lot about a person’s communication style. AI is taking it even further in an attempt to tie these cues to other aspects of job performance.

    The goal of using AI in this way is to solve the age-old struggle of talent identification, which continues to be a challenge for organizations everywhere. Just ask Amazon: they tried to solve this problem with an AI recruiting tool that turned out to be biased against women.

    Critical questions

    Before employers allow voice and facial recognition in their talent selection tools, there are three critical questions to answer.

    1. What’s the connection to job success?

    Experts can train algorithms to recognize anything, including voices, gestures, and facial expressions. In doing so, they have to train the algorithms on what these cues mean, either explicitly or by letting the AI learn through data sets that are fed in.

    Here’s the problem: the market does not have a shared understanding, or even a hypothesis about how these cues are connected to job success. What does it mean if someone looks down, speaks quickly, or pauses to think? Do these behaviors make someone more or less capable of doing the job?

    What if, due to a disability, a candidate doesn’t emote like others do? And what about candidates from different cultural backgrounds, where different expressions mean different things?

    Using AI to help answer these questions is fine, but until there’s a clear job-relatedness link, we’re not ready to deploy algorithms that evaluate candidates in this way.

    2. Will AI reduce bias in hiring, or perpetuate it?

    It’s true that AI does not have an ego or agenda, but that doesn’t make it error-proof. In fact, the biggest advantage of AI – that it’s not influenced by human moods or emotional whims – is also its biggest weakness. Humans, at least, can gut-check each other. AI is completely unaware when outcomes are unfair.

    Also, for AI to learn, it needs humans to tell it what to learn from. We feed data in, and our human biases go in with it. If a training set includes mostly white male faces and voices, for example, then the algorithm will likely favor this demographic. Which could be why Google’s speech recognition is 13% more accurate for men than it is for women.

    So scratch the assumption that AI will free us from bias. Without careful oversight, AI will be just as biased as humans, and on a frighteningly larger scale.

    3. How will candidates react to us data-mining their expressions?

    The best HR tech on the market is not only transformative for internal teams, but it also makes the company look great to the outside world. Sleek, beautifully branded experiences say to the  candidate, “We care about you and want you to feel at ease.”

    This is where video interviewing software shines. Through video, candidates get to showcase themselves in a way that’s not possible on paper or over the phone. And with each interaction, they get to connect with the people and teams they might soon be working with.

    But what happens when candidates find out they’re being evaluated by AI, not on the content of their answers, but on how well they speak and what their faces look like? You can bet this will create more nerves and awkwardness on camera. Some candidates may try to beat the algorithm by playing to what they think the AI is looking for.

    For candidates who agree to complete this type of video interview (because opt-outs will soon become a requirement), they probably won’t show their real, authentic selves. And what good is that in your screening process?

    The right solution

    The HR tech market will continue pushing the bounds, and AI will surely be a player in solving the talent identification problem. But, is voice and facial recognition the right solution?

    It depends on how you plan to use it. If you allow AI to screen out candidates based on a black-box analysis of voice, gestures, and facial expressions, then you could have a moral dilemma on your hands. Do you know how the AI is making decisions and what the adverse impact is?

    While AI may be able to place your next grocery order or recommend a show you’ll love on Netflix, job decisions are a high stakes game. A flaw in the underlying logic of a talent selection tool could derail countless lives, preventing qualified people from getting jobs they deserve.

    Wherever AI takes us, humans will still play an important role in the evaluation of candidates. After all, interviewing exists so that people can get to know each other before working together. AI, then, shouldn’t replace us, but be used to sharpen our skills, scale our efforts, and make our interactions more productive.

  3. Machine Learning vs. Predictive Analytics: What HR Should Know

    If you thought machine learning and predictive analytics were one and the same, you’re not alone.

    Though the terms ‘machine learning‘ and ‘predictive analytics‘ aren’t interchangeable, they are complementary – less like apples to apples and more like apples and caramel. Excellent apart and unstoppable together.

    Food metaphors aside, in today’s hyper-competitive landscape, the only way to get ahead is to be future-ready. To predict things before they happen so you can put your business in a favorable position. Luckily, with all the data that’s available to HR, you don’t have to be a soothsayer to answer questions like:

    • Which candidates in our applicant pool will become top performers?
    • Who on our staff has the greatest growth potential – and who is at risk of turning over?
    • When will I experience my next staffing shortage?
    • What will my expected time to fill be in next year’s economic climate?
    • How will projected business growth effect employee engagement?

    Machine learning and predictive analytics can work together to answer HR’s most burning questions.

    How does it work? Think of predictive analytics as the what and machine learning as the how.

    Predictive analytics is a practice that attempts to quantify possible future events. Predictions are made by finding patterns in current and historical data, often through sophisticated mathematical and statistical models.

    Fun fact: Predictive analytics dates back to World War II when it was used to decode encrypted German messages.

    Machine learning is a way to apply AI to predictive analytics so that predictions can be made without human guidance. Machine learning gives us a superhuman edge because its algorithms can analyze massive amounts of data and identify every possible pattern (and remember – patterns are key to predictions!)

    While predictive analytics can be done without AI, machine learning unlocks new predictive power. Free from the constraints of human analysis, machine learning can use continuous data streams to make real-time predictions, then analyze the outcomes and improve its own performance.

    How can your HR team bring machine learning and predictive analytics into practice?

    As other arms of the business (looking at you, Marketing, Finance…) sharpen their predictive intelligence, the expectation is that HR do same. But where to begin?

    If our peers can predict buying patterns and market trends, then the same capability should be available to HR. And it is! In fact, it’s built into many of the tools that your team is using today, or will be using in the near future.

    Modern pre-employment assessments, for example, are really predictive analytics tools using machine algorithms to identify the highest potential candidates in your applicant pool. You’ll also find predictive analytics and machine learning in resume screening software, recruiting chatbots, and video interviewing platforms, which are beginning to use speech and facial analysis to predict job performance.

    Whatever tools you chose, make sure they’re tied to the questions you’re trying to answer (Who are our high potentials? Where are our risk areas? What resources will we need…?). Implementing predictive analytics or machine learning without a defined purpose won’t do you any good. It can even get you into trouble.

    So go! Explore the brave new world of HR and become the strategic player you’ve always wanted to be. But also, be smart and don’t succumb to everything that’s shiny and new – especially when it comes to AI.

    For help cutting through the hype and investing wisely in new tools…

    Download the HR Buyer’s Guide: How to Evaluate HR Tech in the Machine Learning Era.

  4. HR Technology: How to Avoid Bias

    2019 is a big year of growth for your organization, and your Human Resources department needs help reducing time spent on repetitive tasks so they can focus on recruiting the right candidates and strengthening the team.

    Now for the task of vetting HR software products with buzzwords like machine learning and AI that all make lofty promises. Finding the right solution – whether you’re looking for a chatbot, a video interviewing tool, an onboarding assistant, or a performance management platform – can be a daunting task.

    The good news is, you don’t have to be a software engineer to know the right questions to ask about how these products leverage machine learning and AI. The differences between ‘black box’ and ‘gray box’ technology is one subject of which a high-level understanding will guide you in making the right decision.

    What you need to know about HR technology: ‘Black box’ vs. ‘gray box’ 

    Because machine learning and AI are reliant on the data that goes in to generate answers, even the most sophisticated algorithms are subject to human error and bias. Simply put, if the data going in is biased, the results may be as well (which is what happened with Amazon’s AI recruiting tool in 2015). Here are a few quick tips that will help guide your conversations as you dig deeper into different software providers. 

    Beware of the black box.

    Black box technology is prescriptive and claims it can make decision on behalf of the experts in your department. In a human-first industry like HR, it’s critical to avoid solutions that (1) remove humans from decision-making, and (2) don’t provide transparency into how decisions are made. Considerations like diversity and inclusion may fall to the wayside without human oversight. 

    Stay in the gray.

    Gray box technology is suggestive and delivers recommendations rather than final answers. This keeps humans in the loop and helps avoid unintended bias that can occur in black box technology. Coupling recommendations with insights into the data and what the algorithm was looking for, gray box solutions arm your team with powerful knowledge to make more informed decisions with a personal touch.

    As you begin conversations, remember that utilizing AI and machine learning isn’t a magic bullet capable of solving all your HR problems. A product merely arms the talented people on your team with information to help them make more informed decisions and manage their processes more efficiently.

    To learn more, check out our webinar: How to Keep Your Hiring Process Human in the Age of Automation.

  5. Is Resume Screening Software Biased?

    Resume screening software is a quick and easy way to scan lots of resume data without actually reading resumes.

    The option to automate resume screening – which most talent acquisition leaders say is the most challenging part of recruitment – is enough to perk the ears of any high-volume hiring team. That’s why resume screening software and AI screening software are growing in popularity. But are these tools effective? And do they solve, or perpetuate, bias?

    As companies increase their hiring volume, recruiting teams have to find ways to do more with less. Resume screening is incredibly time-consuming, taking up to 23 hours per hire. And we all know, the longer it takes to screen and hire, the less likely you are to snag a top candidate.

    With so many advancements in AI, why would humans still need to read resumes?

    A better question might be: With so many advancements in AI, why are we still so reliant on resumes? Resumes are problematic for many reasons:

    • They’re self-reported descriptions of work experience and education
    • They include half-truths, exaggerations, and lies of omission
    • They say nothing of knowledge, skills, or character
    • They put too much emphasis years of experience and gaps between jobs
    • Information is hard to verify, because a resume is not an official document
    • Job seekers can easily optimize a resume using keywords, or hire a professional resume writer

    What we’re doing when we use resume screening software is making an ineffective process faster. After news broke of Amazon’s resume screening tool that showed bias against women, this type of AI is under increased scrutiny. But, in Amazon’s story and others like it, it’s not the technology that’s to blame. It’s the underlying data – in this case, the resume.

    A 500-700 word document, even when it contains action verbs and job-related keywords, isn’t a good predictor of success. AI doesn’t change that.

    Luckily, there are other ways to screen candidates at scale:

    • Pre-screening questions are a simpler, lower-tech option. Instead of training AI to scan for keywords on a resume, you could simply ask candidates what is it you want know. Do you have four years or more experience in customer service? You can do this through your ATS, or in your video interviewing platform.
    • A pre-hire assessment integrated into the application process is another good option. This is how American Airlines fuels all their front-line hiring. According to Rob Daugherty, Director of Global Talent Acquisition:

    With a  name like American Airlines, we get a lot of applicants. It’s almost impossible to understand who’s a fit and who isn’t. The assessment helps us focus on candidates with the right personalities and skill sets.”

    • Video interviewing software has also proven its value in time and cost savings. By replacing the phone screen with pre-recorded videos, most companies see at least a 60% reduction in candidate screening time. At Virgin Atlantic, video interviewing enables recruiters to screen 3X more candidates per day.

    AI-driven technologies have unlocked exciting gains in efficiency. What’s important is that we’re driving efficiency in the right areas – not just hiring people faster, but hiring the right people faster. It’s also important that technology is used to inform our decisions, not make our decisions for us. Learn more about AI and the future of hiring in our on-demand webinar: How to Keep Your Hiring Process Human in the Age of Automation.

  6. HR Leaders’ Top 5 Tech Investments for 2019

    HR leaders tell us which of SHRM’s top HR trends they’re most likely to pursue this year.

    Earlier this year, SHRM released the Top HR Tech Trends for 2019. Here are 5 areas where HR leaders are really leaning in:

    AI-Driven Technologies. While new AI technologies are being adopted everyday, HR technology leaders are becoming more diligent about AI tools. There are a lot of exciting applications of AI, and there’s also a lot of hype, so we’re seeing increased scrutiny to test its effectiveness and search for potential bias.

    New Opportunities for Measurement. We’re starting to see organizations take a nontraditional approach to measuring things like employee engagement. New listening-based techniques enable employers to track where employees are spending time, how they’re using internal collaboration networks, and more. This is a big change from three years ago, when 89% of companies were using an enterprise-wide survey to assess engagement, according to Gartner.

    Specialized HRIT Roles. An HRIT specialist is 1.5X more likely to be responsible for data security and technology configuration decisions than IT or functional roles, according to a 2018-2019 HR Systems industry survey. We’re seeing more of these specialists in HR than in finance or marketing because HR deals with more data privacy and integration issues than most other disciplines.

    Specialized Point Solutions. Organizations are showing a renewed interest in technology and innovations from small, emerging vendors. Now that integrations with larger talent management suites can be completed in days or hours instead of months, organizations have the flexibility to implement specialized solutions in recruitment, performance management, and engagement, to name a few.

    Push Recommendations. This trend is all about finding employees at the point of need, whether that need is learning and development content, onboarding information, benefits selection, or something else. This is a great way to deliver customized content at exactly the right moment – an important shift away from formal classroom training and the traditional information dump.

    These are just 5 of the top HR trends reported by SHRM, and we know organizations can’t pursue everything at once. We asked 100 HR leaders in a recent webinar to tell us where they’re most likely to invest this year.

    • 33% said new opportunities for measurement
    • 23% said AI technology
    • 19% said specialized point solutions
    • 16% said push recommendations
    • 9% said specialized HRIT roles

    With so much technology and so many exciting avenues to pursue, HR is in a position to impact the business like never before. The challenge is not, what can we do, but rather, how much can do to drive the business forward? This will depend on alignment with business strategy and smart HR tech investments.

    To learn more about trends, the future of HR, and how to maximize success in Talent Management, watch our on-demand webinar: How to Build a High-Powered HR Machine.