Recently, corporations not only in Korea but also in other countries are adopting Artificial Intelligence (AI) in recruitment. Many of them use AI in document screening, which analyzes the percentage of plagiarism in application forms and the suitability of applicants for their corporations. Some of them even use AI in applicant interviews. Regarding the introduction of AI in recruitment, advocates of it argue that AI ensures fairness and efficiency during recruitment. Nevertheless, there are still a lot of people who resent recruitment through AI. Why do they still doubt AI recruitment? Let us examine the limitations of AI in recruitment one by one.

First of all, AI cannot quantify human capabilities. Many corporations are currently adopting AI in document screening. In order to perform its role, AI must need ‘quantified criteria’ to judge the suitability of applicants. In this case, there is doubt about how AI can quantify its criteria. An applicant who simply uses a lot of words related to a company in his application form may not be a suitable person for that company. In particular, the application form is about ‘who am I’ and ‘what can I do for a company’. To show all the elements mentioned above, each applicant has to describe these elements in their own way. That is, AI should evaluate an individual’s various characteristics and specificity through the application forms. Then, is it possible to evaluate individuals with different characteristics based on the single and quantified criteria of AI? Probably not. For this reason, AI, which requires clear and quantified criteria, cannot accurately determine applicants’ suitability through analysis.

In addition, AI does not guarantee fairness. Some may think that the introduction of AI in recruitment could eradicate corruption, which was controversial in Korean society, as AI analyzes the application forms solely based on data. However, if the data provided to implement the evaluation criteria is biased, no fair judgement will be made. In fact, Amazon, a worldwide corporation in the U.S., had developed AI systems for its recruitment since 2014. However, as AI systems were implemented with gender-biased data (which was a collection of 10 years of accepted applications and most of the applicants were male), AI negatively judged the applications that included words related to women. As the biased result came out, Amazon had to stop developing its AI recruiting system. The case of Amazon shows us two difficulties of adopting AI. To make a fair judgement, clear and quantified criteria are required for AI, and data must not be biased. However, to satisfy all the above, too many considerations are needed in the AI recruitment system.

          Finally, the introduction of AI in recruitment could ultimately lead to generation of uniform human resources. As mentioned earlier, what if AI evaluates the suitability of applicants on criteria like the number of references to a particular word? If it is known to many applicants, they will write the application forms that can get high scores for AI rather than introducing who they are. The same goes for interviews, which determine whether they are able to get along with the members of the company as well as their job competencies. If AI is adopted in recruitment, people will have to pay more attention to their eyes or mouth, which is one of the criteria that is used to judge their facial expressions. What will happen if this continues? Perhaps the uniformity of the evaluation criteria may lead companies to miss talented individuals.

          It seems clear that AI can bring efficiency in recruitment. However, there are still many limitations regarding the ambiguity of the evaluation criteria and fairness for judging suitability with companies that adopt AI in recruitment. Furthermore, there is a potential problem of nurturing uniformed applicants. Finally, we should think about whether efficiency should be a priority in recruitment. Considering that human resources can change the future of a company, it is more important to pursue 'suitable and necessary' applicants for the company rather than 'efficient’ recruitment

 

Conclusion

          It has been a debate about whether it is desirable to introduce AI in recruitment. The supporters of AI in recruitment argued that AI can ensure efficiency and fairness in recruitment and interviewing in person can supplement the limitations. On the other hand, opponents are against adopting AI, claiming that the introduction of AI alone cannot ensure fairness, quantify human capabilities, and could generate uniform human resources. Which one is more efficient, AI's quick and accurate analysis, or judgment only by humans? As the usage of AI is getting wider, this problem will not be limited to recruitment. If people determine the areas of AI and areas where humans can judge more effectively and properly distribute them in developing the recruiting system, satisfactory results for both applicants and recruiters will be made.