A message exchange with an executive recruiter in my professional network brought up a discussion about companies still mainly hiring people based on their CV, i.e. based on their stated experience. Based on what they have done. This is common practice, but that doesn’t mean it’s the right way. The executive recruiter explained: we select candidates based on the facts that are listed on their CV, while we know that it is the person (personality) behind these facts (that are listed in the CV) that determines whether the candidate will be successful. My discussion partner is backed up by empirical studies. For example, one study ranked job experience as only 14th in a list of best ways to select new candidate hires.
The recruiter’s ultimate question was: can we let go of the CV as the most important method for candidate selection?
Why doesn’t the CV provide the right insights?
A CV provides facts (or at times: “wanna be facts”) about what the candidate has done. If the CV is well-written, it will include information about the candidate’s own achievements, as opposed to only a list of roles and dates. Yet we know that this experience often is not what determines the success chance of a candidate.
In her HBR article Experience Doesn’t Predict a New Hire’s Success, Alison Beard explains why experience – as captured in a CV – is not the right way to recruit: “… many measures of experience are pretty basic: the number of jobs you’ve held, tenure at your previous employers, years of total work, whether or not you’ve previously worked in a similar role. Those metrics tell us whether a candidate possesses experience but not about the quality or significance of that experience, which would probably have more bearing on performance”.
What is the alternative to hiring based on CV/experience?
Companies often prefer to hire through referrals (recommendations), because they trust a personal referral more than facts on the CV. So why don’t we always hire based on a referral? The answer is very simple: we do not have enough good candidates in our network to fill all the job vacancies (there are other reasons, e.g. hiring only through referrals would lead to a homogeneous workforce, undermining diversity and innovation; but for the sake of brevity, the first reason suffices).
Let’s have a look at an example demonstrating the weakness of the CV.
In their HBR article Today’s Leaders Need Vulnerability, Not Bravado, Amy C. Edmondson and Tomas Chamorro-Premuzic argue that good leaders – including business leaders – show vulnerability. As a recruiter, you would want this to be a search criterion. Yet how many CVs have you seen where the candidate demonstrated vulnerability on his or her CV (the expected answer is zero)? Recruiters would need tools to find candidates based on such criteria (vulnerability is just an example), using other search mechanisms than browsing through CVs. But how?
Is Artificial Intelligence the future of recruitment?
I believe that the human component is and will remain (at least, for a while) the most important one in recruitment because people – for the time being – trust a personal recommendation better than anything else. And yet, I am convinced that the future of recruitment is Artificial Intelligence (AI). These two statements are not contradicting. While the best selection method remains the personal knowledge of candidates, this will never suffice because the recruiter’s network will never be extensive enough to fill all the vacancies (not by far). And therefore, alternative methods will still be required for most of the job vacancies. I believe that this field will benefit greatly from artificial intelligence.
Let’s look again at the example of vulnerability. While CVs cannot tell you anything about a candidate, one could develop algorithms that learn to detect signs of successfully and constructively demonstrating vulnerability. This is not a straightforward task. It requires careful study of how leaders successfully demonstrate vulnerability, as well as the rationale of vulnerability that is not considered positive. My response to readers who think “this logic cannot be mimicked by software” is that many sceptics who had such thoughts on similar problems were already proven wrong.
Once these algorithms are developed, they can analyze information in the recruiter’s own database, as well as information from external sources. An obvious external source is the Internet, with a focus on LinkedIn, the most important professional social network worldwide. In the future, I believe that more and more external data sources will exist that will offer information on-demand for such algorithms through APIs. Think about platforms such as LinkedIn, Glassdoor and Xing, professional communities (e.g. an association of accountants), social media such as Facebook, Instagram and Twitter (these are never-ending sources of information about a person’s sentiment), loyalty programs and car leasing companies (both hold large amounts of information about corporate users) or even Telecom operators who hold vast amounts of information about business users. Some of these future data sources currently don’t even realize that they could monetize their data in this way. Of course, it all has to fit within the boundaries of privacy legislation.
Back to our discussion about developing AI-based solutions for recruitment. We will probably see a rise of software vendors offering AI-enabled recruitment software tools, using data sources that currently are not used by recruiters, or used only partially. The software vendors’ Intellectual Property will include the definition of the criteria (e.g. leaders showing vulnerability), the algorithms for searching and scoring candidates on these criteria, and the data sources that they use.
Is AI already there?
Some may say that this is not new, and that there already exist AI-based solutions for recruitment. This is true. It’s a developing area. Yet it probably is still in its infancy. Existing AI-based solutions for recruitment support various steps in the recruitment process, such as screening, interviewing, sourcing, applicant tracking, interview scheduling, referrals, job board posting and more. Some of these tools are focused primarily on efficiency gains, i.e. automating the process wherever possible. Other tools use AI-based chatbots for communication. These are not the type of tools that I consider in the current article. The AI that I envision doesn’t aim at efficiency and automation. It aims at implementing intangible, vaguely defined human logic in software, often addressing the soft skills of candidates. I see this as the ultimate challenge of AI, for the time being (once we get closer to realizing this ambition, we will have already set a higher bar).
Does this mean that everything can eventually be automated, and there will be no need for humans anymore? No. In my earlier blog How Human Intelligence Differs From Artificial Intelligence, I argued that computers are good in specialized knowledge, while humans are good in general knowledge. This provides one reason why the ultimate goal is not to automate everything. Another reason is – as explained before – that humans trust humans, and a personal referral is still often seen as the best determinant for selecting candidates. The aim is not to automate everything; the aim is to offer recruiters tools that do a better job than the current tools for the “mass production” component of a recruiter’s job.
Why this is not an easy task
In his HBR article Your Approach to Hiring Is All Wrong, Peter Cappelli writes “… the process moves into the Wild West, where a new industry of vendors offer an astonishing array of smart-sounding tools that claim to predict who will be a good hire. They use voice recognition, body language, clues on social media, and especially machine learning algorithms—everything but tea leaves”. The author explains his doubt: “The big problem with all these new practices is that we don’t know whether they actually produce satisfactory hires.”
I understand this doubt. When technology is a “block box”, and the user doesn’t really know how it works, he/she may not be able to trust it. And sometimes it fails, like human do. After all, software implements logic that humans defined. While understanding this doubt, it does not lead me to the conclusion that the task is impossible. On the contrary, we’ve seen in recent years how technology has been advancing, such that we can now automate tasks that were deemed “impossible” for software in the past. Take as an example text translation. Twenty years ago, this was a massively complex area, and software tools for translation would often be the source for jokes, because the software couldn’t understand the context of text. But current software translation tools are doing an amazing job. Similarly, twenty years ago most people wouldn’t think that driverless cars are feasible, but nowadays this future seems much closer, thanks to advancement in technology and in the ability to capture, formalize and develop software to implement vague and context-rich business logic.
Similarly, I believe that the industry of AI-based recruitment solutions will evolve, and tools will become more intelligent, learning to offer the insights that HR specialists care about. It goes without saying that successfully developing such tools will require tight collaboration between IT people (to develop the software), business analysts, information analysts and domain experts (recruitment specialists). Together, they would have to identify the criteria that software should assess, and formalize the business logic for reasoning with this fuzzy logic. In fact, this is the same process that I did in my PhD thesis. We used the term ontology to describe the (semi)formal representation of the business logic of the unstructured domain of the study (in the case of my PhD thesis, the domain was service bundling; here the domain is recruitment).
Concluding Comment: Removing Bias from Hiring
Although not the topic of today’s discussion, it’s worth mentioning that AI offers another great advantage in hiring/recruitment, namely removing bias. You may remember my earlier blog on workforce discrimination where I reported about a case of a woman of Asian origin in The Netherlands, who only started getting job offers when she adopted her Dutch husband’s surname. In her HBR article Using AI to Eliminate Bias from Hiring, Frida Polli explains two reasons why Artificial Intelligence holds the greatest promise for eliminating bias in hiring: (1) AI can eliminate unconscious human bias; and (2) AI can assess the entire pipeline of candidates rather than forcing time-constrained humans to implement biased processes to shrink the pipeline from the start. Yet another reason why AI is the future of recruitment.
- How to fight workforce discrimination: from “broken windows” to changing a system of beliefs
- Unlikely intersections: How HR practices can stimulate diversity