2nd November 2023
The first workshop took place on June 9, 2023, with 24 participants, and the second on September 29 with 23 participants. Both workshops were managed by five individuals from the University of Iceland and were held at Hannesarholt, a beautiful old house and cultural center located in the city center. The choice of this location was made with the expectation that the environment would encourage productive and creative discussions, which indeed proved to be the case. Participants showed enthusiasm for the topic and expressed strong, sometimes differing opinions on the importance of using AI in recruitment. However, they unanimously agreed on the significance of using AI to eliminate bias. The AI specialists and HR managers often demonstrated a better understanding of the topics discussed, as they possessed more practical experience or had devoted more thought to the use of AI tools in recruitment compared to other participants.
The results of the discussions are summarized below.
The Dangers and Opportunities of Using AI
Despite their interest in incorporating AI into the recruitment process, the participants were cognizant of the potential risks associated with overreliance on AI. Nonetheless, the participants believed that the primary advantage of using AI lies in its ability to process a larger volume of data than the human brain. When used correctly, AI can enhance both the quality and diversity of the recruitment process. It allows for the intentional selection of a more varied group of employees, thus actively seeking individuals with diverse backgrounds and life experiences. AI can introduce entirely new perspectives to the process, although this requires careful programming to identify such candidates, which can be a challenging task.
The participants expressed concerns that applicants may need to create excellent profiles to gain favor with the AI tool, potentially introducing bias as not everyone excels at presenting their qualifications. Moreover, there is a risk of deception, such as applicants typing information in white letters on a white background, falsely claiming to have a degree from MIT, which the AI tool may absorb but human eyes would not detect. As a result, the AI could assign a higher ranking to the applicant based on this deceit, highlighting the need for a critical examination of AI tool outcomes.
To avoid creating more problems than solutions when using AI tools, it is crucial to have a deeper understanding of the technology. Decisions should be made regarding how AI tools can assist in the recruitment process. The tools must be provided with clear instructions about what to consider in applications and what to omit. If diversity is a goal, the AI tool must be informed about what specific characteristics to look for, but this poses a challenge as people from minority groups do not always mention their diversity attributes in job applications, making it difficult for the AI to identify desirable diversity features.
During discussions about the role of AI-based technology in promoting or hindering equality, diversity, and inclusion, the participants highlighted the Icelandic language as a risk factor. Even though 18% of Iceland’s population consists of foreign citizens, most job advertisements are in Icelandic and frequently require proficiency in the language.
What is Important When AI Screens Applicants?
Participants also deliberated on the automatic rejection of applicants from marginalized groups due to unconscious bias. The participants recognized that applicants’ CVs might inadvertently contain information about their backgrounds that could influence hiring decisions. The AI tool must be sensitive to this.
Although the participants agreed on the importance of objectivity and consistency in the recruitment process, they emphasized that no two individuals are ever entirely alike, and defining sameness or similarity with precision is challenging. Therefore, they concluded that a qualitative evaluation is essential and should be transparent. Furthermore, they stressed the need to consider how the applicant fits into the existing team composition, rather than solely checking off boxes. To assess the holistic view of applicants, the AI tool should rely on information regarding workplace policies and the diversity of current employees, while also being able to contact applicants if necessary information is missing from their applications. The AI tool should confirm whether applicants meet the “must-have” requirements for the job and verify the accuracy of the provided information. The participants concurred that the tool should not access applicants’ social media accounts.
In aiming to standardize the recruitment process, the AI tool should ideally draft job advertisements, create interview rubrics, conduct pre-interviews based on application information, and flag potential biases in the process. It was considered particularly important that the AI tool could explain its decisions at all stages of the evaluation, clarifying why an applicant is considered an “excellent” choice compared to others, and identifying differences when not immediately obvious.