CHAPTER ONE: INTRODUCTION

1.0 Introduction
In the last couple of decades, the economy has moved away from the traditional
capital- and labor-based way of working, and toward a knowledge-based way
(Burton-Jones, 2001; Lindmark and Önnevik, 2011). “Knowledge, rather than
labor, materials, or money is becoming the most important resource in the firm.”
(Burton-Jones, 2001, pp. 64). Coherent to this shift, an increased understanding of
which dominant role employees actually play for firms to secure their competitive
advantage has arisen (Lindmark and Önnevik, 2011).

In a knowledge-based economy, the single most important thing for a firm to

achieve its goals and objectives is to have the right people working there, which
directly implies an immense responsibility connected to today’s Human Resource
Management (HRM) departments (Lindmark and Önnevik, 2011). As Chrawshaw
et al. (2014, pp. 4) stated: “An organization is only as strong as its people.”.

1.1. Background of the Study

Artificial Intelligence (AI) is praised for carrying out the most complex tasks with
ease and making various tasks much easier for humans. It is increasingly applied in
innumerable applications, and with growing importance in the Internet of Things
(IoT). Data management has been incredibly simplified by the applications of AI,
and it is used by many organizations to deal with comprehensive statistics, offering
previously unattainable advantages that surpass traditional human efforts, aside
from increasing efficiency in existing operations. One application in recent years is
the deployment of AI to revolutionize the complete recruitment process. AI can
assist organization in various aspects of recruitment, from understanding the body
language of candidates to assisting decision-making. However, there are debates
about the extent to which AI can be effective in recruitment, and in some regards it
has been argued that it is not a viable substitute for conventional recruitment
processes (Belot et al., 2019). Nevertheless, AI increasingly exerts positive
impacts on recruitment, and it is a major asset for HRM and the recruitment
process, especially in achieving equality between the applicants by avoiding
managers’ biased decisions. Therefore, this research investigates and examines the
role of AI and e-recruitment (AIRec) in selecting the right person for the right job
in a company or governmental organization as a case study.
1.2 Statement of the Problem

There is a manifest dearth of investigations of e-recruitment systems and solutions

in developing countries, particularly concerning higher education in Nigeria, where
a traditional and ambiguous process continues to impede national socio-economic
development. The legacy system is based on filling in applications manually, and it
does not allow applicants to track their applications, in addition to the lack of
transparency of how job candidates are selected. Furthermore the lack of
implementations of AI in the recruitment process and applicant quality selection in
a company or governmental organization.

1.3 Aim and Objectives of the Study

Major objectives of the research undertaken in this study are to:

• Investigate the perceptions on the current usage of the e-recruitment system in
Nigerian companies and governmental organizations.
• Implement a suitable measurement survey for e-recruitment technology
acceptance.
• Design and develop an intelligent e-recruitment system with specific needs for
functionalities and services.
• Examine how e-recruitment affects the recruitment quality of employees in
Nigeria public and private sector organizations.
• Evaluate the extent to which AI can enhance the recruitment process of
organizations.
• Evaluate the newly designed AIRec e-recruitment system.

1.4 Significance of the Study

Studies in the field of AI and its implementation in recruitment hold as much

significance as the concept of AI itself, which is ultimately only as useful as its
practical applications. AI is complex, and requires careful management for
effective deployment. While it can enable organizations to explore various aspects
of achieving success in business and deploying the best possible strategies in
employing the new candidates, this is predicated on effective deployment, and there
are innumerable instances of organizations failing in expensive IT investment
projects due to a failure to effectively select and adapt systems and prepare their
organizations for such adoption. However, the efficiency with which the
organizations can handle the tasks for managing employees can itself be explored
by the use of AI (Johansson, et al., 2019).

1.5 Scope of Study

The scope of this study covers Nile University Abuja, Nigeria.

1.6 Limitation of the Study

Financial constraint- Insufficient fund tends to impede the efficiency of the
researcher in applying the standard technical know-how in implementing the
system.

Time constraint- The researcher will simultaneously engage in this study with
other academic work. This consequently will cut down on the time devoted for the
research work.

1.7 Definition of Terms

Artificial Intelligence: the ability of a digital computer or computer-

controlled robot to perform tasks commonly associated with intelligent beings.

Portal: a website or web page providing access or links to other site.

Recruitment: the process of indentifying, attracting, interviewing, selecting,

hiring and employees.

Automation: This is the use of technology or computers to control and process

data reducing the need for human intervention.

Research: A careful study of a subject to discover facts, establish a theory or

develop a plan of action based on the facts discovered.

Software: These are sets of logically related instruction given to the computer to
perform specific tasks.
Analysis: The person that is entrusted with the job of systems analysis in an
organization, the same as system analyst

Record: Is a collection of related fields, a record is one complete set of fields.

Information: These are processed data.

Program: A set of instruction given to the computer to use in processing

information.

Input: The data given to the computer to use in processing information.

Files: A collection of related records.

 CHAPTER TWO: LITERATURE REVIEW

2.1 Human Resource Management and The Recruitment Process

2.1.1 Human Resource Management

Human Resource Management (HRM) is a management function focusing on
hiring, motivating, and maintaining an organization’s workforce. It can be defined
as “a distinctive approach to employment management which seeks to achieve
competitive advantage through the strategic deployment of a highly committed and
capable workforce using an array of cultural, structural, and personnel techniques”
(Crawshaw et al, 2014, pp. 7).

One of the most common models used in HRM research is The Fombrun Model,
see figure 1, which defines the main operations of HRM and their interconnections
(Fombrun, 1984). The main operations are Selection, Performance, Appraisal,
Development, and Rewards (Lindmark and Önnevik, 2011; Fombrun, 1984). Each
of these operations has different functions and processes associated with them, all
developed to ensure a return of investment connected to the labor force working in
organizations (Crawshaw et al., 2014).
 Figure 1. The Fombrun Model of HRM (Fombrun et al., 1984).
The first component of the Fombrun Model, defined as Selection in figure 1, is about how
organizations work with the recruitment and selection of people for different positions and job
openings. Following the purpose and delimitations of this thesis, the Selection component is the
one that this study focuses on. The other components are about HRM processes connected to
stages where the people affected are already part of the organizations, and therefore lie outside
the scope of this work. In more recent research, the Selection component
is re-defined as “The Recruitment Process”, which therefore is the definition used in the
remaining parts of this thesis.

2.1.2 The Recruitment Process: An Overview

The recruitment process focuses on how an organization is working to recruit and select people
for different positions (Lindmark and Önnevik, 2011). It can be defined as “a process of creating
a group of qualified candidates for the vacancies within the organization”
(Stoilkovska et al., 2015, pp. 284). To understand the scale of the recruitment process, and its
different stages, an overview of the overall process flow can be seen in figure 2.
Figure 2. The stages of recruitment and selection (Bratton and Gold, 2012).

Job Analysis and Internal/External Factors

The process can be said to begin with the job analysis, which together with internal
and external factors, impact the workforce planning (Bratton and Gold, 2012).
Internal factors can for instance be a high rate of absence, product development,
and development of new services. External factors, on the other hand, can be
aspects such as increased demand for products and services, new laws or
regulations, and new kinds of technology (Lindmark and Önnevik, 2011). After an
organization has analyzed its current position based on aspects such as these,
together with its current employment situation, it is time to map out the needs that
the organization has for meeting challenges and possibilities lying ahead
(Lindmark and Önnevik, 2011; Bratton and Gold, 2012).

Workforce Planning
By applying various research techniques, the organization determines what kind of
knowledge, skills, abilities, and other characteristics and competences are required
to meet said needs (Crawshaw et al., 2014; Bratton and Gold, 2012). After
analyzing this, the decision if to start new recruitment(s) is taken, and if so,
whether it should be done internally or externally (Crawshaw et al., 2014;
Andersson et al., 2016).

Recruitment
If the organization decides recruitment is needed, the first step is to decide on
selection criterions which can come from defining what a job role involves, and
what would make an effective worker in this role (Crawshaw et al., 2014; Bratton
and Gold, 2012). The organization must be clear on which competence(s) are of
the highest value for the job and which ones come second (Andersson et al., 2016).
Following the definition of selection criterions, choices on which selection
assessment technique(s) to use need to be made (Crawshaw et al., 2014; Andersson
et al., 2016). Measures of reliability and validity of different selection assessments
should help guide choices about which technique(s) to use (Crawshaw et al., 2014,
pp. 144).

Establishing a time frame for the process in this stage and appointing
responsibilities connected to the process in the organization is favorable (Bratton
and Gold, 2012; Breaugh, 2008). A strategic goal is to minimize recruitment costs
and parallel to this, shorten the time of recruitment to reduce the loss from empty
position(s) (Stoilkovska et al., 2015).

Subsequent to this, the organization needs to work out a strategy for recruitment
activities to carry out in order to advertise and communicate the job vacancy, and
attract job applicants by describing tasks, personal requirements, benefits, and
other information (Breaugh, 2008; Crawshaw et al., 2014). Decisions on suitable
methods need to be taken, and if parts of the process will be outsourced. In the case of
outsourcing, quality assurance of external actors needs to be done (Bratton and Gold, 2012;
Breaugh, 2008).

Applicant pool
Data about the people applying for the vacancy are collected through the selected
tools and create a so-called applicant pool consisting of potential candidates
(Lindmark and Önnevik, 2011).

Selection
When having the application pool, the selection part of the process begins, usually
starting with a screening of the applications based on the selected criterions and
with the help of assessment techniques chosen in earlier steps of the recruitment
process (Bratton and Gold, 2012; Thebe and Van der Waldth, 2014). This narrows
down the application pool before potentially suitable candidates are assessed more
carefully to determine who should be selected (Crawshaw et al., 2014).

Job Performance
The selection of a candidate to hire does not necessarily mean the end of the
recruitment process. As seen in figure 2, the final step is “Job Performance”, which
entails performance tracking and career development. Without a future follow-up
of the candidate, it is impossible to evaluate the success of the recruitment (Kerrin
and Kettley, 2003, Alan et al., 2016). This part of the recruitment process lies
outside the scope of this thesis.

2.1.3 The Recruitment Process: Challenges

Candidate Shortage
One major challenge for organizations connected to the recruitment process is
attracting and retaining skilled employees, commonly specified as talent
management. Increased competition in finding and recruiting skilled workers
intensifies this challenge even more (Alan et al., 2016). As mentioned in the
introduction section of this thesis, a study conducted by the Confederation of
Swedish Enterprise shows that the main problem connected to recruitment in
Sweden 2021/2022 is the increased difficulty organizations have to recruit people
with the right competences. As many as seven out of ten organizations experience
difficulties recruiting the right people, and three out of ten recruitment attempts fail
entirely due to a lack of candidates (Svenskt Näringsliv, 2022).
Another challenging aspect connected to the competition for skilled employees is
the time-consuming parts of the recruitment process. Due to the candidate
shortage, and the courtship candidates with commonly requested competences get,
many of them are involved in several application processes parallel. Subsequently,
this puts pressure on organizations to have fast recruitment processes. If they are
slow, the risk of the candidate choosing to move forward with another organization
is immense (Crawshaw et al., 2014).
Potential candidates’ views on organizations, i.e. employer branding, is also
playing a vital part when the so-called “war for talent” intensifies following the
candidate shortage. If the selection, screening, and feedback process connected to
recruitment have been a negative experience for candidates, the chances of them
applying to the organization again are limited.
Therefore, it is relevant to ensure that all parts of the recruitment process are
efficient and effortless to increase the chances of candidates applying again.
Connected to this, the technologies used in the recruitment process should be user-
friendly, and at the same time, preferably also cut costs connected to recruitment
for the organization (Van Esch et al., 2019).
Objectivity
Another commonly mentioned challenge in recruitment is to ensure objectivity
when selecting between potential candidate(s) and reduce the impact of biases
associated with subjective assessment of candidates (Crawshaw et al., 2014).
According to Crawshaw et al.
(2014), there are three main effects following biases in perception that all people
are subjective to as human beings. These are: The Halo effect, The Similarity
effect, and the Beautiysm effect. The Halo effect implies the tendency assessors
have to generalize a good or a poor rating of one of a candidate’s attributes to all
attributes of said candidate. The Similarity effect points to the fact that assessors
tend to rate candidates whom they perceive as similar to themselves higher than the
ones they do not. The Beautiysm effect implies the tendency assessors have rating
candidates they perceive to be beautiful more highly than those they do not
(Crawshaw et al., 2014). The three effects together account for close to 62 percent
of the variations in assessors’ subjective ratings of candidates (Crawshaw et al.,
2014).

Technological Development
One challenge, which in many ways also can be an opportunity connected to the
recruitment process, is technological developments. Technology is a primary factor
in shaping the way people work, and adopting the appropriate technology to
promote effective and efficient processes is crucial for organizations’ survival
(Crawshaw et al., 2014).

Okolie and Irabor (2017) have concluded a literature study on the potential
challenges of e-recruitment, which can be said to be the latest big technological
breakthrough in recruitment pre artificial intelligence. From this can be seen that
some crucial potential challenges of technological developments are privacy and
data security connected to the information gathered about applicants following its
implementation. Therefore, before adopting any kind of online recruitment
method, impacts, strengths, and weaknesses of said methods should have been
carefully studied (Okolie and Irabor, 2017; Omolawal, 2015).

Another challenge connected to technological developments is the quality and

quantity of candidates applying. In getting broader exposure with the help of
technology, many organizations have reported getting a large number of
applications from unqualified people (Bhupendra and Swati, 2015; Omolawal,
2015). This creates additional work for recruitment staff members who then must
review more resumes and more emails, and need to install expensive software to
track numerous applications (Armstrong, 2009). Screening large amounts of
resumes, checking skill mapping, and the authenticity of applications are all time-
consuming tasks for organizations (Anderson, 2003; Mathis and Jackson, 2006).

Lack of knowledge about technological developments within the HRM

community, as well as limited commitment to them by senior managers are also
potential challenges (Armstrong, 2009; Omolawal, 2015). From a survey on the
use of e-recruitment conducted by Kerrin and Kettley (2003), cultural and
behavioral change is reported as the most significant challenge in ensuring that
new technology delivers.
In addition to this, how potential candidates think about technology developments
could also be a challenge. The more technological solutions used in the recruitment
process, the more computer savvy is needed from applicants (Bhupendra and
Swati, 2015). Also, the limited personal interactions during the recruitment process
that follows the implementation of e-recruitment are leading to frustration on the
applicants' part due to missed opportunities to share information about themselves,
and gather additional information about their potential employer (Robertson and
Smith, 2001).

2.2 Artificial Intelligence (AI)

The term Artificial Intelligence (AI) can be seen as an umbrella term that defines
and indicates the implicit intelligence of machines (Kaplan and Haeinlein, 2019a).
In broad terms, an AI can be defined as an intelligent system with the ability to
think and learn (Jarrahi, 2018). The system can be thought of as one that mimics
general human abilities such as processing information, learning, problem-solving,
and speech, which makes it behave like an intelligent human (Jarrahi, 2018;
Russell and Norvig, 2010). Elaborating further, AI can be explained as a system
that can interpret and learn from large quantities of data to accomplish stated goals
by adapting to situations it exists in (Kaplan and Haenlein, 2019a). Artificial
intelligence is relevant anywhere an intellectual task is supposed to be performed,
making it a truly universal field (Russel and Norvig, 2010).

Even though artificial intelligence, and research about it, have been around since
WWII, and the term first was stated by John McCarthy already in 1955 (McCarthy,
2007; Russel and Norvig, 2010), the concept is still blurry to most people, and
many still wonder what AI actually is and what it does (Kaplan and Haenlein,
2019a).

According to Kaplan and Haenlein (2019b), there are three main reasons why it is
challenging to define AI. The first reason is the difficulty in defining human
intelligence to start with, which in turn complicates the connection of intelligence
to a machine (Kaplan and Haenlein, 2019b). The concept of intelligence is
discussed, and the difficulty of finding a universal definition implicates effects on
the understanding of artificial intelligence (McCarthy, 2007). The second reason
why the concept of artificial intelligence is still blurry to people, according to
Kaplan and Haenlein (2019b), is what is referred to as the “AI effect”.

The AI effect implicates the concept that once people have gotten used to a
machine completing a certain complex task, the ability to do this task is not seen as
a sign of intelligence anymore. This causes the definition of AI to become a
moving target, and by this always seeming to be out of reach (Kaplan and
Haenlein, 2019b). The third reason is the many different evolutionary stages AI
has, going from narrow to general to super intelligence.
On top of this, based on its cognitive, emotional, and social competences, AI can
be classified into analytical, human-inspired, and humanized (see Table 1). The
different kinds of stages and types are easily mixed up, which confuses the term
itself (Kaplan and Haenlein, 2019a).

Types of AI Systems
As mentioned in the paragraph above, AI can be classified into three groups based
on its cognitive, emotional, and social competences; analytical, human-inspired,
and humanized (Kaplan and Haenlein, 2019a). As can be seen in Table 1,
analytical AI only has cognitive intelligence meaning that it generates a cognitive
representation of the world and makes future decisions based on past experience.
Analytical AI can for instance be used in image recognition. Human-inspired AI on
the other hand, has both cognitive and emotional intelligence, giving it the ability
to recognize and analyze human emotions. As an example, this kind of AI system
can be used to identify peoples’ emotions over video calls. Lastly, Humanized AI
has all three competences; cognitive, emotional, and social. An AI system of this
kind could for instance be a virtual agent with real-time customer engagement
(Kaplan and Haenlein, 2019a).

This classification is mainly for understanding how different kinds of intelligence

can be seen in AI systems, but can also be linked to conscious and unconscious
human behavior which in turn lays the foundation for the data being used in the
learning processes of the AI systems.

Cognitive intelligence can be argued to be a more straightforward kind of

intelligence that humans are conscious of. Emotional and social intelligence on the
other hand is something that is more unconsciously built up through the different
contexts humans are parts of (Kaplan and Haenlein, 2019a). However, from a
technological viewpoint, the defining aspect of all three types of AI systems is
their ability to learn from past data which in broad terms can be done through three
types of learning processes; supervised learning, unsupervised learning, and
reinforcement learning (Kaplan and Haenlein, 2019a). More about these learning
processes in the following section.

Table 1. Types of AI Systems (Kaplan and Haenlein, 2019a).

Analytical AI Human- Humanized Human

inspired AI AI Beings
Cognitive    
Intelligence
Emotional x   
Intelligence
Social x X  
Intelligence

Artistic x X x 
Creativity

Supervised Learning, Unsupervised Learning, Reinforcement Learning

How does AI work?

Artificial intelligence can be explained as a sort of classification technology.
Fundamentally, it can be said that the primary function of the technology is to
differentiate and classify data. Simplified, AI agents are trained on (large) datasets
with the aim to learn how new data input should be classified (Russel and Norvig,
2010). The fundamental necessities to have a simple, working AI are, therefore:
large quantities of data relevant for the task it is meant to solve, software with
programmed algorithms advanced enough to draw conclusions from said data, and
lastly, it must be run on hardware with the ability to process and store the data
efficiently (Grosan and Abraham, 2011; Savola and Troque, 2019).

Machine Learning
Machine Learning (ML) is a subcategory to artificial intelligence with the
approach that the algorithm is learning to improve itself through past experience
and data (Russell and Norvig, 2010; Kumar, 2018). One example of a machine
learning application is Natural Language
Processing (NLP), which entails the process where AI systems understand and
analyze language used by humans (Jarrahi, 2018).

Machine learning is divided into three main types depending on which kind of
feedback being carried out; supervised learning, unsupervised learning, and
reinforcement learning (Russell and Norvig, 2010; Kumar, 2018). Supervised
learning entails that input-output pairs are provided to the agent, and it learns to
map an input to the correct output. To exemplify, an input can be an image of a cat,
and the corresponding output would then be “that’s a cat”. In unsupervised
learning, the agent learns patterns in the data without being provided with explicit
feedback on what is “right”. The most common unsupervised task is clustering,
where the AI agent clusters similar data points gradually developing concepts. For
instance, a taxi agent could develop a concept of “good traffic days” and “bad
traffic days”, without ever being given labeled examples of each. Reinforcement
learning encompasses learning where the agent learns by rewards and punishments,
often by applying points to actions based on how favorable the action is. For
example, ten points for a win at the end of a chess game tells the agent it did
something right (Russell and Norvid, 2010). Often, a combination of the different
kinds of feedback is used.

Deep Learning
Deep Learning (DL) simply explained, is a branch of AI that as closely as possible
tries to mimic how the human brain works, in the sense that it learns from
experience. If machine learning is a subset of artificial intelligence, deep learning
can be explained as a subset of machine learning (Russell and Norvig, 2010), see
Figure 3. Deep learning focuses on the use of neural networks and algorithms used
to train neural networks to learn from large volumes of data. Neural networks can
be explained as a group of mathematical models aiming to mimic how the human
brain works. The deep part of the term comes from the fact that neural networks
are built up by several hidden layers, and to be classified as deep learning, at least
two layers are required, giving so-called multilayered neural networks (Campesato,
2020).

Figure 3. Artificial Intelligence, Machine Learning, and Deep Learning (Dhande,

2017).

As can be seen in Table 1, every type of learning (Supervised, Unsupervised, and

Reinforcement) can be connected to each different type of AI System (Analytical,
Human-Inspired, and Humanized). In other words, AI solutions (including its sub-
categories ML, and DL) can have different types of cognitive, emotional and social
intelligence following how they are developed.

2.3 Artificial Intelligence in Recruitment

2.3.1 The Trend of AI in Recruitment Today
Artificial intelligence is believed to be a powerful tool with the ability to bring new
opportunities for both organizations, businesses, and services and,
correspondingly, contribute to the evolution of new skills and working methods. It
is believed to be a key catalyst for innovation and development (Regeringskansliet,
2019).

According to Reilly (2018), recruitment is one of the organizational functions that

is most likely to embrace and undergo considerable innovation and transformation
in the near future, and as of 2018, a trend among recruiters in adopting AI software
and solutions was observed (Upadhyay and Khandelwal, 2018). Since then, AI has
become one of the most spoken of trends in the recruitment industry (Upadhyay
and Khandelwal, 2018).

One essential reason why AI is making its way into the recruitment industry is the
fact that the competition for skilled employees is getting more intensified. For an
organization to shorten its time-to-hire does not just represent an efficiency gain,
but also a strategic advantage in the battle for human capital (Van Esch et al.,
2019; Faerber et al., 2003). These advantages are followed by the fact that AI can
process information and make decisions at volumes and speed far exceeding
human capacity (Van Esch et al., 2019). Implementing AI into a recruitment
process will also facilitate potential candidates’ access, which further will increase
competitiveness for the organizations doing so (Hmoud and Laszlo, 2019).

Despite AI solutions’ promising features and potential gains, and the increasing
demand for them, AI tools are not yet used to a great extent in the recruitment field
(Upadhyay and Khandelwal, 2018). A lot of organizations are showing interest, but
due to the fact that most organizations are new to AI solutions, a lack of
knowledge slows down AI's entry into the market (Van Esch et al., 2019).

Even though organizations using AI are not the benchmark in the recruitment field
yet, the use of AI tools is becoming progressively more frequent. The more
organizations using AI, the more pressure will fall upon other organizations and
HRM managers to adapt and incorporate AI into their recruitment strategies to not
become outdated (Hmoud and Laszlo, 2019). According to a study made by Okolie
and Irabor (2017), HRM professionals in organizations using AI solutions in their
recruitment state benefits including reduced costs, increased number of applicants,
better candidate matching, easier application processes for candidates with a wider
variety of job opportunities, and also a greater response rate on receiving feedback
from applicants. In the longer run, the use of AI has the ability to not only change
organizations internally but make an impact on the whole recruitment industry by
affecting revenue, profitability, and talent acquisition (Savola and Troque, 2019).

2.3.2 Different Types of AI and Automation Tools

Today, AI tools for almost every step of the recruitment process can be found on
the market, starting to transform the recruitment field more innovatively (Sekhri
and Cheema, 2019).
Pre-screening applicants’ CVs to detect and evaluate keywords to match
candidates to appropriate job openings through Applicant Tracking Systems (ATS)
is one of the more common ones (Upadhyay and Khandelwal, 2018). AI-powered
chatbots are also getting popular in recruitment processes. By using natural
language processing (NLP), chatbots can process information and interact with
candidates and enable real-time engagement through text messages, email, and
social media, to mention a few communication ways (Upadhyay and Khandelwal,
2018; Nawaz and Gomes, 2019). Another AI-powered tool making its way to the
recruitment industry is video chat analysis. These can be programmed to analyze
candidates’ features such as age, tone of voice, cadence, word choices, mood,
behavior, eye contact, emotions, etc (Fernández and Fernández, 2019). Following
the large volumes of data existing connected to people on social media nowadays,
another AI feature in the recruitment field is to scan social media spaces. Both to
find the right candidates, and to assess candidates’ social values, beliefs, and
attitudes to get an understanding of their personal and professional traits
(Upadhyay and Khandelwal, 2018).
In a study made by Savola and Troque (2019), a table of the evolution of
recruitment stages is presented to indicate how the recruitment industry works
today, and how it can come to work moving forward, see Table 2.
 Recruitment Stages Current Tools Next-Generation Tools
Sourcing & Screening Broad-based tools: Social networks and data
- Applicant tracking driven platforms:
systems (ATS) - Facebook
- Job boards - LinkedIn
- Chatbots
Applicant tracking & - ATS - AI candidate
Interview scheduling - Customer Relationship correspondence
Management (CRM) tracking
tools - Self-scheduling tools
- Traditional interviews
- Scheduling tools
Assessment - Skill-assessment tools - AI-based
are available psychometrics and
- Manual evaluation behavioral assessment
tools
- Video interviewing
- Solutions and analytics
based cultural fit
assessment tools
Background checks, - ATS - Automated background
Offer management & - CRM and onboarding checks
Onboarding tools that can be - Automated offers
integrated to ATS - Chatbots that answer
basic HR questions

Table 2. Recruitment evolution (Savola and Troque, 2019).

2.3.3 The Viability of Using AI in Recruitment

In a systematic literature review done by Javed and Brishti (2020) on “The
viability of AI-based recruitment process”, four themes in literature about AI in
recruitment are identified, namely: Actor Role, AI Role, AI Adoption, and
Potential Risks. This section will follow the same structure and is therefore split
into four sub-section, one for each theme.
[Link] Actor Role
The actor role emphasizes on the actors, both recruiters and applicants, and on the
advantages and opportunities they will be able to achieve from the implementation
of AI in recruitment (Javed and Brishti, 2020).

According to Savola and Troque (2019), and Hmoud and Laszlo (2019)
administrative and repetitive tasks will be replaced by AI technologies and
following this, the recruiters' job tasks will shift more toward strategic functions
and leadership roles. Upadhyay and Khandelwal (2018) mean that even if AI will
take over a lot of activities in the recruitment process, humans are still needed
when it comes to building relationships with candidates and to put focus on talent
identification and development. Following AI’s way into recruitment,
Kaplan and Haenlein (2019b) write that it will be important for HRM managers to
incrementally adopt leadership styles suitable for organizations in transformation.

Bhalgat (2019), Upadhyay and Khandelwal (2018), and Savola and Troque (2019)
all state that, even with AI opening up new ways of working with recruitment,
humans will still be needed in the recruitment process. Upadhyay and Khandelwal
(2018) mean that human recruiters are needed to interpret and quiz candidates
when it comes to understanding their emphatical and emotional qualities. Savola
and Troque (2019) argue that AI will be used to bring forward the best potential
candidates, but humans will be needed to make the final hiring decision.

Srirang et al. (2020) and Nawas and Gomez (2020) argue that the candidates’ view
on the use of AI tools in recruitment plays a vital role for how successful the use of
AI will be. Srirang et al. (2020) bring forward that it is important to consider any
concerns candidates might have regarding exclusion, discriminiation and privacy,
and find ways to assure them about the neutrality of the recruitment process and all
tools being parts of it. Nawas and Gomez (2020) mean that one way to do this is to
show openness and transparency connected to the algorithms used in the AI
solutions, and by doing so, the organizations will show that they play fair.

More and more candidates are open to engaging with AI-based systems according
to VanEsch et al. (2019). They argue that this comes from the systems being
perceived as novel, convenient and empowering by the candidates. Another reason
stated by Nawaz and Gomez (2020) could be that for instance, an AI-powered
chatbot could help candidates in their progress by enabling real-time responding to
questions, general communication, and feedback. Another thing mentioned by Van
Esch et al. (2019) that makes AI systems in recruitment positive for candidates,
and at the same time employers, is that by giving candidates a smoother
recruitment process with ongoing interaction throughout it, the likelihood of
rejected candidates applying to other opportunities in the same organization is way
higher than if candidates did not had this kind of experience.

However, Javed and Brishti (2020) stress that these prominent advantages for
actors in the recruitment process following the use of AI cannot be achieved until
the actors understand what AI is and how its work could benefit them.

[Link] AI Role
The AI role emphasizes on the impact of AI in recruitment as a powerful
technological tool. Mainly focusing on the dominant advantages of it (Javed and
Brishti, 2020).

Bhalgat (2019) and Van Esch et al. (2019) mean that the most noted advantage of
using AI in recruitment is its ability to minimize human biases, and being more
objective than humans.

Michailidis (2018) states that the selection of candidates will be more diversified
using AI than being done by humans, due to AI’s possibility to reduce biases and
maximize objectivity.
According to Van Esch et al. (2019), this advantage comes from that AI can
process information, and make decisions based on said information, at volumes
and speed far exceeding human capacity.

By adding AI, and thus minimizing human biases, Bhalgat (2019) means another
level of transparency is added to the process, which will lead to an improved
attitude toward organizations using AI among potential candidates, and that their
employer branding will be improved following that.
From a study done at Unilever, which Van Esch et al. (2019) highlight, it was
reported that by using AI, both the speed and quality of their recruitment process
increased dramatically. AI’s ability to speed up the recruitment process
significantly, while not compromising on its quality, is what Upadhyay and
Khandelwal (2018) mean justifies the way AI will change the recruiter role. Savola
and Troque (2019) proclaim that AI will replace humans in the initial stages of the
recruitment process, such as candidate sourcing and attraction, selection, screening,
and testing, and allow for human recruiters to only focus on the later stages.
According to Geetha and Reddy (2018) AI helps to identify potential candidates’
skills, competence, and experience, and match these factors to the job they are
applying for, resulting in the hiring of talented candidates.