In this article we’ll investigate the pros and cons with AI in recruiting and whether it’s a good idea or not to start using AI in your day-to-day. Let’s kick this off with a short introduction to AI in recruitment.
If you’re not already familiar with AI tools for recruiting, please become acquainted with the help of our Ultimate Guide to AI Recruiting. Whether you’re middle management, in HR or business in general, understanding this technology is of great benefit to you since it is the future of many businesses.
What does AI in recruitment entail?
Have you ever, in your daily work life, worked on a task that you find monotone or unnecessarily time-consuming? We dare to assume you have and with that said, we’ve got good news for you; there is an AI tool for many of those tasks. As of today, the most proof-tested usages of AI in HR involve tools for time and cost-efficiency. To mention a few areas in which these tools are being used:
- Screening
- Sourcing
- Automatic Shortlisting
- Candidate Database Management
- AI resume parsing
If you’re curious as to how each tool functions, read up on it here.
You’re most likely aware of the whole “to be, or not to be”-debate regarding AI. And it is a debate for good reason, because we are yet to be fully knowledgeable in this, its potential dangers and negative outcomes. However, the tools mentioned above are machine learning-based applications and are more often used in recruiting because they are designed to support humans in their decision making by;
- analyzing text
- finding patterns and
- comparing data.
The less commonly used tools are mainly those that are ADM-systems (Automated Decision Making). The reason is that ADM-systems are more complex and designed for autonomous decision making. ADM-systems, in comparison to ML-based applications, compare statistics based on chosen factors and then provides a result. These two types of AI could potentially be damaging for people depending on what kind of data and biases it’s been fed with. Something we’ve written about here: Why Amazon’s AI-driven High Volume Hiring Project Failed.
Now that you know a little bit about how AI is used within the recruitment sector we’ll seek to get an understanding of the different skillsets of humans and artificial intelligence.
What humans are better at today (and how it might change)
Empathy, creativity and independent thought are three pillars of the human experience and thereby three key factors that differentiate us from other animals, as well as from AI.
A great manager is great because of her emotional intelligence. A great scientist has a knack for gathering a lot of information but at the same time makes use of her creativity in order to reach conclusions no one has ever made before. One trait of a great recruiter is her ability to keep an open mind and not discard candidates just from a first glance.
In a work environment these are all invaluable skills, but we have another important player, one that happens to be; love. However fluffy that may sound, we are loving creatures and in a work environment, having supportive, understanding and up-lifting colleagues are of great importance for our performance. We’re not talking about romantic love but rather companionate love; when you feel connection and affection. On top of that, it’s difficult for a human to feel empowered by a machine in the same sense that another human can single-handedly increase a person’s confidence with some positive feedback.
Therefore, soft skills are difficult for AI to attain and since love, connection, and empathy are invaluable sources of good for people, the idea of having an AI as your HR manager might feel far-fetched. Whilst having the AI assist you in managing the performance of people is another story.
Apart from the above, humans are also excelling in human communication, language-wise. Natural Language Processing (NLP) is a branch of AI with the objective of interpreting, deciphering and understanding human languages.
As of today, it’s still extremely complex to teach an AI all of the subtle, implicit meanings that a single spoken sentence can hold. Humor, intention and delivery (body-language and tone of voice) all play their part in how a message is received in between humans.
Not to talk about connotations. “Feeling blue” is a good example of a tricky connotation to teach an AI. How will the AI know that we’re feeling sad and not blue as in the color blue? How would the AI know that someone is lying during an interview? Socially intelligent humans are experts in noticing even the slightest shift in a persons’ being and from that draw conclusions that she’s either lying, covering something up, or just doesn’t feel comfortable enough to share the truth with you. This all boils down to an important factor in an interview-situation; a person's vibe.
It’s not important to an AI that a candidate gets along well with it, whilst it is very important for the team that the candidate is a good cultural fit . Surely, the AI can run personality tests and gain an understanding of who she might be as a person, compare those to the colleagues personality types and eventually reach a conclusion of whether she’s a good match or not for the company. But it is hard to truly assess the match without physically meeting in person.
Recruiters typically have a good understanding of the people they’re hiring for and a skilled recruiter can take into account more than just personality traits and skill-sets, you can see subtleties in a person’s personality, the “vibe” if you so like. You can also decipher whether a candidate would get along well with the other employees. Could an AI ever reach that same type of underlying, emotional intelligence? And is it even necessary to have that ability in order to make a good recruit?
What AI is better at today (and how it might change)
In his book The Globotics Upheaval, Richard Baldwin suggests that the entrance of AI into this world will disrupt it more than globalisation, industrialism and automation did. It’s a bold statement to make but most likely true given how AI beats us to it when it comes to; processing information at rapid speed, analyzing big data and even at providing more accurate medical diagnosis. It makes AI incomparable to a human cost-wise for companies.
To give you an example; the DBS bank in Asia relieves 40h/month from their employees by using AI driven recruitment for monotonous tasks and time-to-fill. AI is inexorably more time-efficient as well as more accurate when processing information, what it all boils down to is the humans dedication to teach the AI. Because even if humans are innately capable of, and skilled at, interpreting languages, differentiating objects and recognizing images, it’s also something AI’s can learn.
The difference is that machines require an exhaustive dataset to learn and mimic those skills which will take time, but will it take more time than birthing a human and have it go through school? No. Point being that, as of today, AI excels in everything that includes quickly processing and analyzing big data, for example; scanning through 3000 job applications only to provide you with the top 10 candidates (merit-wise), twenty minutes later. It is also, if programmed right, better at being unbiased than humans (a hot topic for anyone hoping for AI to deliver non-biased hires). This is indisputably great in order to diminish discrimination in workplaces but it’s a long way coming till we can rely solely on AI to make completely unbiased hiring decisions. The AI has to have, in that case, been taught with as little human influence as possible since it is the human influence which creates bias in the AI.
It’s with this idea that other problems occur though, if we extract humans out of the equation at the same time as we build AI for humans, what are we left with and would it be dangerous?
What experts have to say on the topic
In a canvassing of 979 experts in the AI field, they were asked if the increasing impact of AI in our world would leave humans better or worse off in 2030. The canvassing showed that 63% of the respondents thought we would be better off whereas 37% thought we will not be. This canvassing was regarding AI in general but it gives some clues as to how much impact AI should be allowed to have in our daily lives with the conclusion; “A number of the thought leaders who participated in this canvassing said humans’ expanding reliance on technological systems will only go well if close attention is paid to how these tools, platforms and networks are engineered, distributed and updated.” This leads one to believe that having an AI as a recruiter could be beneficial considering time- and cost-efficiency as well as a probable diminishing of biases but only if carefully surveilled and programmed into minute precision.
Examples of professions that have been replaced by AI
Online today you mainly find examples of professions that most likely will be replaced by AI, such as; telemarketers, receptionists, couriers and proofreaders. This is certainly valuable to have an understanding of in terms of how you could direct people away from certain jobs and industries. But it’s also interesting, perhaps even more tangible, to look at the world today and play with the thought that for every visible machine, there used to be a human. This in return gives a hint as to what is to come. Take ATM’s for example, for every ATM there used to be several bank clerks. Cashiers are soon also a thing of the past as self-checkout desks are becoming more common and the store personnel will soon only be there to assist when the machines are being problematic. We mentioned the uprising of AI in medicine earlier, and despite the importance of a trusting doctor-patient communication as well as a doctor’s bedside manners, there are now robots performing surgery, handing out medicines and suggesting treatments.
This is a perfect example of AI and human collaboration which leads us to a conclusion.
So, will AI replace recruiters?
No. Not right now but maybe in a far off future. Because just as in the doctor-patient example, when it comes to empathy and human understanding, humans are undoubtedly the experts. Empathy is a gateway to human connection and connection in itself equals trust. If you were to be employed solely by a robot without ever having met a colleague at your new job - how would it make you feel? Perhaps you would trust that the AI made the right decision regarding you but wouldn’t there be something missing in your joy of getting the job? A sense of disconnection to your co-workers maybe, or an indication of falsity. Point being, if we are to rely solely on robots and thereby disregard the human connection, we as humans would have to change our behaviours and what is important to us in a workplace. Thereby completely disregard our innate need for connection and affection.
Let’s move into the future and think of a world in which both humans and robots operate, the only future which we can accept is a world in which the two live in symbiosis with each other. AI can never be given the upper-hand in any given circumstance, simply because humans have to be in charge in a human world or else all dystopian movies on the subject might come true.
