AI and Recruiting – threat or opportunity?

Posted on June 6, 2019 by Addy Lee
Article image - AI and recruiting

Artificial Intelligence is a hot topic today in research. With applications mushrooming, scientists and entrepreneurs are beginning to make sense of finding a place for AI in recruitment now…

Artificial Intelligence could be generally categorized into three levels: weak, medium or strong. Weak AI represents bots like AlphaGo, which simply obeys the rules set by humans. Medium AI is designed to mimic the human-reasoning process and use it as a guide. Strong AI is something you can see in the movies, which should closely resemble the current level of human intelligence.

The current technology can only rise to the medium level, but it’s still valuable for many applications. After training, Medium AI could handle defined and repetitive tasks, and allow humans to focus on the more demanding aspects.

“The weakest level of artificial intelligence was developed in the 1990s, like IBM’s Deep Blue Master Chess Player. While there are still a few technologies making use of this variety of AI, most fall solidly into the middle tier. The ‘middle’ level of AI is where technology currently stands. Machine learning uses the human reasoning process as a guide, but does not fundamentally replicate it.”


Whether it’s a corporation, a contingent or executive search firm, reach and time-effective resume screening remain the greatest challenges in recruitment. 52% of recruiters said that the most difficult part of recruitment is funneling suitable candidates from too many large candidate pools. How to improve efficiency, while containing time and cost, becomes important. Google finds recruiters usually have no choice but to use Crtl+F (search), searching the keywords of a JD to screen resumes, which to be honest, is both time-consuming and inefficient.

Now, there are several training process for AI, which can enable it to efficiently assist you on sourcing and screening. AI could rapidly exhaust the talent pool and external sources, and produce a ranking list based on the degree of matching. The process could be dramatically speeded-up. Unilever employs AI technology to recruit employees. The time spent by recruiters in examining applications has decreased by 75%. Xiao Bing, an intelligent robot developed by Microsoft, served as an ‘interviewer’ for the interns of the AI group of the Internet Engineering Institute for Microsoft (Asia). In just 12 hours, Xiao Bing completed the preliminary screening of the interview and selected more than 3500 candidates from more than 12,000 people to enter the next recruitment stage!

AI is transforming to a new model.

Now many companies are training an AI model, which could digitize CVs and JDs into E-CV and E-JD, and then simply compare their degrees of match, including personality, technical skills, education, background, etc…

To train this AI, we only need to input articles, books, languages, and internet sources. Therefore, the bias and complexity could be reduced and minimized. The model could be used for general purpose selection.

Apart from that, the existing voice recognizers and chatbots can support better candidate communication. The aim of first calls by contingent firms is simply laying out the basic job descriptions and asking for a YES or NO, which chatbots could easily handle. Therefore, it greatly reduces the work pressure on researchers, and saves time.

What if somebody told you that AI might reinvent executive search?

Executive search firms have quite different value adds compared with contingent recruitment models. The breadth and depth of their advice oftentimes are what clients truly want and need, to perform. That being the case, AI’s relevance may seem better suited to volume recruitment purposes.

Having said that, it doesn’t prevent AI to be involved as a support role in multiple stages during the executive search process. The capability of sourcing and screening could dramatically release the valuable time of researchers from simple and repetitive tasks. Furthermore, based on our research results, some AI solutions providers, such as HireVue and e-Cheng, can provide a digital summary of candidate’s performance during the interview, in order to avoid any unexpected mistakes or biases. Although it still has many areas to improve, the tool is valuable for researchers. The scored numerical reports could help researchers give clients a clearer picture of fit, plus consistent a presentation that helps.

Our submission is therefore:

  1. Users should carefully select the relevant AI model. Different models have different strengths and weaknesses. There is no TRUE or FALSE in this decision, simply FIT or LESS FIT.
  2. Firms need to understand or at least be familiar with the algorithms behind AI. It is not simply an input/output black box, but an efficiency enhancer, which allows us to screen big loads of resumes, otherwise it’s not feasible. It is important to find key words that match the particular roles in question; plus find ways to overcome the limitations of the database deployed.
  3. Preventing Bias is the very important. Last year, Amazon had to roll-back a good AI-enabled selection system due to bias issues.
  4. Firms may need to be clear that, since AI has not been fully developed, you cannot rely on it singularly. It is exciting if AI could help significantly,
  5. but human inputs & interventions make things perfect.

“AI can possibly revolutionize HR, but we need deploy with care.”


To conclude, AI has promising possibilities to improve the current process of executive search for both fit and speed. Having said that, right now they are still a work-in-progress that require sensible human-tweaking. So, is AI a threat or opportunity to recruitment currently? I believe you have the answer now…