Hiring is often a challenge for most organizations. From candidate sourcing and screening to talent assessment and offer management, every stage of the process has the potential to create a time lag, which in turn can hurt the business. Manual screening of resumes is still the most time-consuming part in the recruiting process, especially when half of the applications received for a job are unqualified.

These days, some companies have chosen to explore the fields of artificial intelligence (AI) and machine learning to find the best talent. This article questions the effectiveness of AI and whether it can help with screening and hiring decisions.

70% of HR managers believe the recruitment process would be more effective if it was more data-driven (Source: Recruiting Times)

AI Can Learn Human Biases

AI providers aim to reduce unconscious bias and to make objective hiring decisions. This is great news! However, AI is not an independent intelligence form. It is a system trained by humans with criteria and data provided through human intervention. This means that any human biases within the recruiting process could easily be picked up by the AI software as well.

In a famous experiment, recruiters reviewed identical resumes and tended to select more applicants with white-sounding names than with black-sounding ones (Bertrand, Chugh, & Mullainathan, 2005). If AI software learns what a good hire “looks” like based on dirty data, it will make biased hiring decisions – just like humans. This illustrates how AI can misinterpret input data and make workplaces just as homogeneous as they were before.

AI technology companies promise to find the best talent (Source: Mobe)

Assessments are less biased than AI technology as they ignore irrelevant information such as a candidate’s age, gender, and race and include only relevant information such as cognitive ability, personality, and skills, which have been shown to predict job success through years of research. They offer a decisive advantage over AI as candidates are assessed along the dimensions that really matter for successful job performance (Tracey, Sturman, & Tews, 2007). Having the same input data allows for a fair comparison and ranking of candidates.

Lack of Human Touch

Certain situations require human intervention, explaining the “human” in Human Resources. When ambiguity or contextual concerns arise, they often require a level of organizational, social and emotional intelligence that only humans can provide. SuperCareer utilizes resume information as well as personality and ability tests to screen applicants for an open position down to the best fitting candidates in ranked order. At this stage of the hiring process, assessments are more objective when compared to individuals or AI technology for these decisions.

SuperCareer acknowledges that hiring decisions should be guided by an algorithm-informed individual, rather than the other way around. The company offers proven methods to screen resumes and evaluate applicants based on predictive assessments. It provides recruitment services by industry experts to ensure that the selection process benefits from vital human involvement. This combination draws on the best of both worlds, encompassing key pieces of information from structured interviews that feed into the algorithm to provide the best match for candidates to jobs.

AI software analyzing facial expressions in a video interview (Source: Affectiva)

The Black Box at the Heart of AI

Recruitment tasks like sourcing, screening can be performed by AI technology. In fact, it might even be possible that some day AI technology might take over the entire hiring process. However, it is important to understand how machines learn these processes. The background mechanisms of AI technology are still obscure. Before we trust and rely on AI technology for hiring, we must learn the inner workings of the black box that is at the heart of AI technology.

In conclusion, like many other industries, HR is going to be data-driven. AI promises to help and we are sure it has a bright future. However, we doubt that this will happen in the immediate future. We must strive to eliminate factors that contaminate recruitment and selection decisions and use only information that strongly correlates with successful job performance and organizational needs. We also think that good AI platforms will probably be assessment based. As of now, this is the direction that SuperCareer is taking. We hope to add value to the hiring process and establish credibility to algorithmic talent matching with our approach.

What do AI models really measure? (Source: llrx)


Lara Herzogenrath
Mehmet Civan
Suman Kalra
Arkin Kora


Bertrand, M., Chugh, D., & Mullainathan, S. (2005). Implicit discrimination. American Economic Review, 94-98.

Tracey, J. B., Sturman, M. C., & Tews, M. J. (2007). Ability versus personality: Factors that predict employee job performance. Cornell Hotel and Restaurant Administration Quarterly, 48(3), 313-322.