In the 12 months before May 2022, 78.4 million people changed jobs across the US. Coordinating a task of this magnitude (including recruiting, onboarding, training, and retaining that sheer number of people) presents a major human capital challenge. In fact, there simply aren’t enough people in the hiring, HR, and recruitment industries to cope with this level of movement, at least not without the assistance of tech automation.
But when it comes to using technology to aid hiring, there is a rational fear that as much as it assists in the process, what if it is in fact hindering as well? How do we actually know we can trust AI to make the right choices? This is precisely what I’m going to discuss here. And spoiler alert: the answer is more nuanced than you might think.
The short answer is that yes, of course, technology is good for hiring and the reality is we could not cope with the demand of candidate applications without it. According to Zippia, on average, there are 250 applications for each open job position. If you multiply that times the 78.4 million people who changed jobs in the US this past year alone (roughly twice the population of California), it becomes clear very quickly why technology is not only good, but in fact necessary for hiring. It simply wouldn’t be possible to process that many people into companies at the required pace without AI, platforms like applicant tracking systems (ATS), AI chatbots, job boards, and the many other alternative niche platforms that exist.
However, these technologies are not without their flaws. So assuming that we can all agree that technology is not only good for hiring, but in fact necessary, it is important to also talk about the main flaws. The good news is that we are aware of them and are currently working to actively improve the core technologies underpinning our hiring experience. Let’s discuss some of the main flaws below and some possible solutions as well.
AI Can Be Gamified:
This is absolutely true. As a job seeker, you can locate a role you wish to apply for and in order to maximize your chances of getting an interview with a human, simply take the job title and description (including any obviously prominent keywords), add those to your resume and submit it. Then sit back, relax, and watch how quickly the AI believes you are relevant and passes you through the first stages of matching.
This is a heavy vulnerability point for AI matching technology and those who know this weakness often use it to make their resume relevant. This is frequently used by candidates looking to get through applicant tracking systems that use plug-in matching software. The AI is not intelligent enough to realize that the jobseeker copied components of the job description in their resume, which is something that a human would be able to spot immediately. This highlights some of the issues regarding how intelligent technology (such as AI) actually is in the recruitment process. I would argue that currently it’s primitive but effective at coping with the sheer volume.
We also see the same issue with AI chatbots. Often an applicant can quickly establish the correct answers to give in order to pass through the screening. Again, a human interviewer would quickly be able to identify and ask further clarifying questions to prevent and screen out any inaccurate claims or false answers from a candidate.
AI Struggles with Transferable Skills:
Typically, AI is designed to prioritize the job title and the hard skills in the body of a job description by cross-referencing them with the resume. Against the masses, this generally does work effectively. However, as anyone who has ever worked in recruitment knows, transferable skills are often a wonderful way of discovering talent, especially when there is a shortage. AI is generally very literal and looking for an exact match so it would reject the candidate, while a human recruiter would be able to spot the hidden potential.
AI Has Bias:
Depending on the layer of machine learning that has been applied to AI to make it smarter, this has been proven to be true. But how in fact can AI gain a bias? It’s actually very easy. AI itself does not care about your race, gender, or class—what it cares about is looking at data and finding trends which it believes will increase the likelihood of relevance and a match. Let’s say you have matching software that is machine learning from a data set of IT resumes—a generic pool of maybe 10,000 software engineers. Well, statistically that data is likely to have 80% male resumes (over half of them will also be white) since that is currently the gender and race trend within that talent market.
These candidates will naturally have similar trends on their resumes, such as masculine language, certain extracurricular activities, and general queues that the AI would use to identify that this resume looks remarkably like the types of candidates that are most common in the industry. Because of this unintended bias, the AI will often end up showing a preference for white males’ resumes in the hiring process. Not only has this been proven, but in fact Amazon, who spent many years developing its own AI matching software for the purpose of hiring, famously scrapped it in 2018 after determining it had learned a masculine language bias.
What I find most concerning about this is that there are many platforms and software out on the market that offer AI matching in order to provide the best shortlist of candidates for a job. How many of them have been using machine learning with an unintended bias that actively promotes further bias within the hiring process? It truly is a terrifying thought to reflect on the technology burst of the 21st century and how we have so quickly built websites, software, and apps without truly forecasting or even really understanding the mass-scale impact they may have on society.
The job market is no different—what have we really been doing with AI when it comes to helping promote diversity? Now, there are companies with billions of dollars invested in them that are heavily focused on building technology that caters to and addresses this bias, which is fantastic, but what does it say about the potential damage that may have already been done?
Technology Can Be Frustrating for the User Experience:
Technology has created many layers of filtration and while this is good for the mass-scale funneling of candidates, for the jobseeker, the process can feel very much like a hurdle of confusing user experiences. You might start your journey on Google, land on a job aggregator, then get redirected to a job board, which in turn sends you to an applicant tracking system (ATS). Keep in mind that you’ll have to sign up for all three layers in the process and that you don’t even apply for the job at the company you want until you reach the ATS. At that point, you’ll enter into the “black hole” of waiting since in many cases the AI behind the ATS will not actually tell you whether your resume was successfully matched to the job. Even if you take the initiative to follow up and see if your application was matched, there’s no guarantee that you’ll receive a response or ever find out where you are in the hiring process.
During the above process, you may also be faced with a chatbot that has a very direct set of questions you’ll have to answer without the context a human recruiter would be able to provide. The upside is that the interview with the chatbot is an opportunity to “sell” your skill set and relevance for the job. Overall, the hiring process (despite technology) is still a very fragmented and inefficient process. It sorely lacks the strong streamlined processes we have for user experiences with other digital ones—for example, retail has Amazon, logistics has Uber, and takeout has DoorDash.
Summary:
While this article may feel like it has simply been bashing HR tech, I can assure you the benefits do outweigh the above-named issues. Even though there are many companies out there working to fix these issues, we still have a lot of work to do and it must remain our focus. The future of work demands that the above and many other areas are resolved in order to provide a more streamlined, inclusive, and efficient onboarding process.
At the end of the day, the technology that has been created for hiring is good, but it’s certainly primitive in comparison to other verticals. There is a lot of work to be done, but the good news is that our digital society is providing us with the data required to build better, more intuitive technology. And as long as we remain aware of these flaws and limitations we can and will find solutions.
Author: Arran Stewart, CEO, Job.com
Original Article Published by:
출처 : Korea IT Times(http://www.koreaittimes.com)