How to use AI responsibly in Recruitment

For many, AI conjures up images of Hal from 2001: A Space Odyssey. It’s a good idea in theory but a dystopian nightmare in practice. But despite our apocalyptic fears, AI is strangely marketed as a recruiter’s new best friend. In fact, I can’t help but wonder that if they’d named it Augmented Intelligence instead of Artificial Intelligence, it would have been widely adopted already.

Alas, plenty of recruitment software companies like to shout about their “cutting-edge AI” and “machine learning algorithms”—something that we mere mortals will never understand—and it seems to work: 96% of senior HR professionals believe that AI can greatly leverage talent acquisition and retention.

But what exactly is it? And what implications does using AI in recruitment have for your company? In this post, we will be deconstructing the term “Artificial Intelligence” and looking at the benefits (and limitations) of AI recruiting in 2020.

What sort of AI is used in recruiting?

First, it’s important to know that AI is a general term that covers several types of machine learning. Technopedia succinctly defines AI as “the creation of intelligent machines that work and react like humans.”

AI is machine learning that mimics human intelligence such as decision-making, visual and speech recognition and translation.

But how does it work? Put very simply, AI is made up of complex algorithms that analyse large amounts of data, find patterns and then make determinations based on prediction. What’s more, these algorithms don’t need constant tweaking by human coders, they improve themselves based on evolving data!

The large quantities of applicant data gathered in the recruitment process are a natural resource for AI. Especially if you have a high staff turnover or a large volume of applicants for a specific role. For example, most applicant tracking systems track candidates through the recruitment process. Then, attempt to match their skills to job roles. But this is just the beginning. Read on to see all the different ways you can implement elements of AI in your recruitment process.

How to use AI in recruiting

Candidate sourcing

AI can save a lot of time by automatically scouring the web for social media profiles, publications, and other search engine results that may provide insight into which candidates are suitable for and receptive to new opportunities.

When AI is used on walled platforms such as job boards and social media sites, that’s fine. However, many third-party tools now claim to do this from outside platforms using “publicly available data”. There are question marks over the legality and morality of using these tools, with many arguing that it breaches personal data rights.

Scheduling interviews

Scheduling interviews is a simple but time-consuming task that most people would gladly offload to AI. You can purchase clever tools like Calendly or Doodle outside of your ATS, but hopefully, your ATS has an interview scheduling feature already built-in.

Candidate screening

Research suggests that up to 75% of CVs are under-qualified. Sifting through them is time-consuming. Having an algorithm to pre-select only the most suited candidates will save you time in evaluating those with the right skills and qualifications more rigorously. 

You can use AI in different ways here. It can use data from unsolicited information, such as social media profiles or facial movements on video calls, to identify suitability. We recommend avoiding these types of solutions unless you want a lawsuit.

You can also use it to identify job suitability from pre-authorised, job-relevant data such as resumes or pre-hire assessment scores. For more information on candidate screening tools, check out our guide to pre-hire assessments.

Candidate re-discovery

Did you have suitable applicants in your last cycle that didn’t get the job? Candidate rediscovery is particularly useful for people who recruit at scale or employ seasonal workers. AI can be used to analyse your talent pool and recommend who might be available when you’re next looking to hire. AI-powered Talent CRMs or ATSs can send timely updates and alert you to candidate behaviour on social media sites such as LinkedIn.

Diversity hiring

Unconscious bias is unavoidable in hiring. However, you can reduce it using fair and objectively scored pre-hire assessments instead of manual CV reviews. The role of AI isn’t significant in reducing bias; in fact, many AI solutions can perpetuate biases if trained on pre-existing “biased” hiring data. 

White-box, transparent assessment solutions like ThriveMap’s Realistic Job Assessments can provide you with an audit trail of how each candidate was scored against the job-relevant attributes you identified in your ideal candidate profiles. Black-box AI solutions can’t provide this transparency, which can lead to candidates taking legal action against you.

Replicating “human” contact

One of the functions of AI in recruiting is that it can react in real time to language. Some companies have used this technological development to create “chatbots”,; the little chat bubbles that appear on many career pages asking applicants if they need help. Many chatbot providers claim applicants are more likely to complete an application if they have a chatbot to guide them through the process. There has, however, been some public backlash on chatbots being used to dehumanise the hiring process.

Interview analysis

Recent advances in AI technology include the ability to analyse facial expressions, spoken language and body language of applicants. This is particularly relevant given the increased dependency on virtual interviews and the more general movement towards remote hiring. This relatively new technology isn’t commonly adopted in hiring because of legitimate concerns about cultural differences and context not being picked up in the algorithms. Still, it’s one to keep an eye on.

The limitations of AI

It requires lots of data.

The amount of data needed for an algorithm to assess a candidate’s suitability for a single job role is huge. Before employing AI in your recruitment process, consider the diversity of your job roles and the volume of applicants. Candidate selection algorithms will have a better payoff if you’re regularly hiring many people for the same position.

It’s not totally unbiased.

While AI may help to remove unconscious human bias from your recruitment process, it’s not a silver bullet. Because machine learning algorithms work by finding pre-existing patterns from analysing biometric data from existing employees, the AI may unknowingly mimic any existing biases. AI finds patterns, not causes. You’ll also need real human intelligence to ensure your hiring is fair and inclusive.

It can make mistakes

One of AI’s biggest appeals is that it almost eliminates the possibility of human error. However, there’s always a chance of error in the data that the algorithm uses to find patterns, learn, and improve. If this happens, it can be very difficult to correct the error after the AI has already “learned” from it.

Candidate scepticism

Ever since AI’s implementation into business, there have been fears that robots will take over our jobs; think of self-checkouts and driverless trains. While the reality is that more jobs are created than are lost with technological advances, this fear of AI hampers its implementation rate. This fear extends to candidates who may want to know why they have not been selected for an interview.

Is it actually “artificial intelligence”?

The futuristic, sci-fi connotations of artificial intelligence mean it’s become something of a buzzword in recruiting, when in fact, very little of the technology and software used is really artificial intelligence. Matt Alder of the excellent Recruiting Future podcast, says, “We don’t actually have any genuine AI in recruiting.” This isn’t to say that software that matches applicants to jobs isn’t useful; it is just that it isn’t necessarily AI. 

So beware of recruiters and software companies that sell their services on their “cutting-edge AI” because it’s likely that the marketing is ahead of the science.

In closing…

It’s all too easy to misunderstand AI’s role in recruiting. Before implementing AI into your process, it’s worth determining what tools are best for your business. 

Automation is a good option if you’re a volume recruiter looking to reduce your time-to-hire. However, rather than getting caught up on the solution you need, think deeply about the problem instead. 

AI is not a silver bullet that can fix all your problems. Recruitment is about people, and (at least now) we can’t rely on technology to do it all for us.