AI and the Recruitment Process: Issues and Perceptions in the Tunisian Context

AI and the Recruitment Process: Issues and Perceptions in the Tunisian Context

Sofiane Founès, Sami Boudabbous

University of Sfax, Faculty of Economics and Management, Tunisia

DOI: https://doi.org/10.51584/IJRIAS.2025.100800063

Received: 06 August 2025; Accepted: 13 August 2025; Published: 10 September 2025

ABSTRACT

Through semi-structured interviews and thematic analysis, we uncovered some important insights about the digital maturity of HR professionals in our region. While many participants showed a theoretical understanding of AI, their practical application of these technologies was surprisingly limited. It’s interesting to note that while tools like LinkedIn are being used and contain AI features, many users don’t recognize these elements and often mix up automation with real AI capabilities. The findings reveal that, for most, AI is mainly seen as a way to automate repetitive administrative tasks rather than a valuable resource for assessing soft skills and cultural alignment. There are ongoing concerns about algorithmic bias, the standardization of profiles, data privacy, and overall transparency—which are valid issues that need addressing.

Moreover, we found that economic factors are a significant barrier as well, especially given the high costs of implementing AI technologies. This is especially relevant for many businesses in Tunisia that are facing challenges with low recruitment volumes. One of the most interesting insights from our study is the connection between the number of people that companies are hiring and their opinions on how effective AI is. This relationship really stands out as a key takeaway.

Therefore, in this study we emphasize the importance of digital transformation for the HR sector in Tunisia. This is not simply a trend but essential in terms of how one stays relevant and what organizations and their employees are looking for. Getting over this transformative can be beneficial for better engagement, efficiency and success for all the stakeholders.

We should see AI as a supportive tool that complements the efforts of human resources professionals, rather than viewing it as a replacement for the essential human touch that is so critical in this field.

Keywords: Artificial intelligence, recruitment, human resources, digital transformation.

INTRODUCTION

At present, the professional landscape is going through a rapid change due to the advent of increasingly high-performing technologies and an ever-shifting socioeconomic environment (Kvint & Bodrunov, 2022). Which in this context tunisians companies must not be indifferent on the changes transforming the world’s structure and organization. Globalization, the explosion of data, the increasing digitization of services, and the new demands of emerging generations are forcing organizations to rethink their operating methods, particularly in the strategic area of human resources management (HRM) (Shaddiq & Irpan, 2023).

AI is now emerging as one of the major drivers of this transformation (Horowitz et al., 2022), and in order to capitalize on these changes, they need to not only adopt these practices, but also anticipate their effects. Artificial Intelligence (AI), which has been any of various systems that can simulate human-like processes including learning, reasoning, or decision-making, has become associated with organizational innovation (Gjølstad & Skants, 2023). In Tunisia, more and more companies, particularly in the banking, technology, and service sectors, are beginning to integrate AI solutions into their internal processes, including those related to human resources. Jarrahi (2018) emphasizes that these technologies not only enable the effective handling of substantial volumes of data, but also for the generation of faster and sometimes more objective decisions than those made by humans.

According to some authors (Benabou et al., 2024; Gupta, 2024; Huang et al., 2023), the impact of AI on the HR function is multiple: automation of administrative tasks, improvement of the candidate experience, personalization of training, and, above all, transformation of recruitment methods. Roppelt et al. (2024), state that artificial intelligence (AI) is viewed as a fundamental tool in talent search and selection, allowing for an efficient identification of profiles aligned with business demands (Evalato, 2024).

Nevertheless, the implementation of AI in HR processes definitely raises questions, even concerns, especially in respect to the effect on employment (Suhonen, 2025). This debate over whether machines will displace humans is a hot topic in Tunisia as it is elsewhere.. The debate over the replacement of humans by machines is heated, in Tunisia as elsewhere.

In Tunisia, this dual observation is particularly relevant, given the structural unemployment affecting many graduates, particularly in managerial disciplines. One of the most important questions in this situation is how hiring managers feel about incorporating AI into their procedures. This question assumes particular importance in a nation such as Tunisia, where employment reforms occasionally fall behind technological advancements. Increasing efficiency while maintaining fairness and the human element of the hiring process is a dual challenge for Tunisian recruiters. This requires a detailed understanding of both the benefits and limitations of AI in this field.

To provide answers to this question, we opted for a qualitative approach based on the deductive method. Using this strategy, Tunisian recruiters from a range of industries participated in semi-structured interviews. This approach was chosen in order to collect detailed and nuanced testimonies, which enabled us to determine attitudes, anticipations, and concerns about the application of AI in the hiring process.

Our work will be structured around three main sections. First, a literature review will highlight the main research studies on the interaction between AI and the recruitment function. The second part will then present our methodological approach, explaining our data collection and analysis choices. The empirical findings from our interviews, which show recruiters’ concrete experiences and perceptions of AI, will be presented in the third section..

LITERATURE REVIEW

1.1. AI

1.1.1. Genesis and historical trajectory of AI

AI itself is not a new idea, but the last few years have witnessed an unprecedented distillation of AI development and applications, especially in organizations (Lu, 2019). Although its gaining increasing interest now, it can be traced a few decades back (Daugherty & Wilson, 2024). Alan Turing, a well-known British mathematician, who, in 1950, published a seminal paper called “Computing Machinery and Intelligence,” laid one of the first conceptual landmarks by asking whether machines will ever be able to think (Kuipers & Prasad, 2022). In it, he proposed a protocol that became famous as the Turing Test, aimed at determining whether a machine can imitate human intelligence to the point of deceiving a human interlocutor. However, Turing did not explicitly use the term AI; This only appeared in 1956, at the Dartmouth Conference, thanks to John McCarthy, who defined AI as the science and engineering of manufacturing intelligent machines, including brilliant computer programs (Gonçalves, 2023). However, Raj and Kos (2023) claim that although this act lays the foundation for AI, it was not until the 1980s that AI could be seen as an applied technological field. AI was at that time still in its early days but initial industrial applications focused around automating repetitive tasks in factories, especially robotics.

By contrast, hard proof of AI did not reach the general public until the mid-1990s, for instance during the high profile matches where IBM’s Deep Blue program defeats world chess champion Garry Kasparov (Papajorgji & Moskowitz, 2024).

This is an unprecedented event that marks the shift in the flavour of collective consciousness on AI — From an intellectual curiosity, to an almost blanket reverence for a technology with orders of magnitude greater capacity to perform certain (complex) tasks.

Since then, the explosion in the volume of data generated daily, advances in computing power, and the widespread availability of the internet have provided fertile ground for the spread of AI (Zhu et al., 2023). Artificial Intelligence (AI) is now used in many industries, including finance, logistics, human resources management, healthcare, and agriculture, thanks to these major technological advancements (Elkholy et al., 2025).

1.1.2. Conceptual complexity and definition of AI

Along with the propagation of AI, there has been an attempt to clarify the concepts; however, ironically, there are a plethora of definitions reflecting the inherently abstract nature of AI germane to the field. Others — like Bidyalakshmi et al. (2024) classify AI as an algorithmic, cognitive, and learning-based systems that can perform missions commonly reserved for human intelligence. This definition emphasises perception, data processing and analysis, logical reasoning, and the independent decision-making capability.

Conceptual diversity reflects the plurality of coexisting theoretical approaches. Some authors, such as Jha et al. (2023), associate AI with the automation of cognitive processes such as problem-solving or learning. Others like Korteling et al. Zhu et al (2021) define it as the ability of machines to perform tasks that have historically only been performed by human beings. An agent is defined as anyone able to perceive and reason in and out of the environment (Dumais, 2023). In this vein, few researchers propose a pragmatic notion, for instance Natale (2021), points out that AI should be “the technologies which are able to reproduce human functions such as seeing, listening, understanding speech, and making decisions”. This multitude of views reveals a key truth: AI is not a stagnant definition, but an ever-evolving domain, continually redefined.

A fundamental difference between what researchers refer to as “weak” and “strong” AI is revealed by this complexity (Bory et al., 2024). Systems that specialize in particular tasks but do not claim general intelligence or self-awareness are referred to as having weak artificial intelligence (Li et al., 2025). This type of AI is now largely dominant and present in many applications: search engines, voice assistants, recommendation systems and even automated recruitment software (Tiwaari et al., 2025). Conversely, strong AI, still hypothetical, refers to the potential emergence of machines capable of transferring their intelligence to a wide variety of contexts, or even developing a form of artificial consciousness (Seth, 2024). For Bellaby (2024), this distinction fuels scientific and ethical debates, because it raises fundamental issues about the very nature of intelligence.

1.1.3. Technical paradigms, contemporary applications, and ethical issues

The evolution of AI can be understood through different paradigms that have marked the history of the field. Initially, symbolic AI, dominant between the 1950s and 1980s, was based on modeling human intelligence through logical rules and symbol manipulation (Augusto, 2021). This trend, represented in particular by expert systems, proved effective in controlled environments but limited in dynamic and ambiguous contexts. Faced with these limitations, connectionist AI emerged, inspired by the functioning of the human brain and based on artificial neural networks (Bardozzo et al., 2024). According to Vassallo et al. (2024), this approach, based on learning from data, enabled considerable advances in the 2000s thanks to the development of deep learning, a technology that now forms the backbone of many AI applications. Finally, statistical AI, which has dominated the field over the past decade, relies on massive data processing using algorithms that, without any real understanding, are able to make predictions with increasing accuracy (Kodieswari et al., 2024).

These technological advances have enabled the proliferation of concrete applications of AI. Automatic natural language processing, conversational agents (chatbots), visual recognition, and predictive tools are now ubiquitous (Ghosh et al., 2024). According to Dandotiya et al. (2024), AI is a multifaceted technology, capable of rapid adaptation in contexts as diverse as personalized medicine, cybersecurity, and education. As techniques improve, the vocabulary of AI also expands, encompassing concepts such as supervised, unsupervised, and reinforcement learning, semantic analysis, and multi-agent systems.

But this technological advancement does not come without raising serious questions, particularly ethical ones. Several authors, such as Priyadarshini & Cotton (2021), insist that only technologies capable of truly autonomous self-improvement should be considered intelligent. From this perspective, current systems remain limited, despite their apparent performance. Other voices, such as Díaz-Rodríguez et al. (2023), call for ethical governance of intelligent systems, emphasizing the need to ensure their transparency, fairness, and accountability. Debates surrounding algorithmic bias, data protection, and automated decision-making in sensitive areas such as recruitment and justice serve as a reminder of the extent to which AI raises major political and social issues.

More broadly, the evolution of AI questions our relationship with technology and its effects on social structures. Taking a more critical approach, the theory of technological determinism suggests that technical advances profoundly influence our modes of organization and communication. While some authors, such as Johnson & Wetmore (2021), defend a “hard” view of this determinism, believing that society is shaped by technology, others (Symons & Abumusab, 2024) prefer a more nuanced approach, recognizing the ability of social actors to guide the uses of AI according to ethical and political choices.

Given this historical trajectory, theoretical complexity, and diversity of applications, AI appears less as a simple technical innovation than as an ongoing cognitive and social revolution. It invites a deep reflection not only on what machines can do, but also on what we expect from them in a constantly changing world.

1.2. The recruitment process

In a context of rapid transformation in the world of work, the recruitment function now occupies a strategic position within organizations. Once perceived as a purely administrative and support activity, the role of the recruiter is now being reconsidered in light of the growing challenges related to talent management (Kwon & Jang, 2022). This evolution is explained by the increased recognition of human capital as a determining factor in organizational performance. As Hadijah (2024) points out, one of a company’s most decisive assets lies in its ability to identify, attract, and retain the right talent. This ability is no longer solely a matter of internal human resources management, but constitutes a major lever for competitiveness in an increasingly tight labor market.

According to Vardi & Collings (2023), the notion of talent here goes beyond simple technical skills; it encompasses a set of individual attributes, including knowledge, intelligence, experience, and also personality traits. However, in many contexts, companies are encountering increasing difficulties in accessing these rare profiles. Several cyclical and structural factors explain this dynamic. On the one hand, the mismatch between the skills acquired by graduates and actual market needs, sometimes referred to as a skills gap, fuels this tension (Tulgan, 2015). This phenomenon, exacerbated by initial training sometimes deemed insufficiently adapted, constitutes an obstacle to hiring in many sectors, particularly technical and digital ones (Morandini et al., 2023). On the other hand, workers’ expectations have undergone profound changes. In particular, we observe a growing desire to balance work and personal life, greater mobility, and a less stable relationship with work than in the past (Fasang & Aisenbrey, 2022).

Demographics constitute another explanatory factor. The aging of the workforce, coupled with the gradual departure of baby boomers and a decline in birth rates, particularly in developed countries, is significantly reducing the pool of skilled labor (Hetrick et al., 2021). This situation is reversing the traditional balance of power between employers and candidates. Now, it is often the latter who hold the negotiating power. They are able to select companies based on their values, their salary offers, but also their ability to provide a fulfilling work environment. This means that employers no longer simply ask candidates what they can bring to the company, but also, in return, present the benefits they are able to offer them (Dobbin & Kalev, 2022).

These changes have contributed to intensifying the “war for talent,” a concept widely documented in the literature since the late 1990s (Kwon & Jang, 2022). While the skills shortage is not a new phenomenon, it has worsened to the point of becoming a central issue for organizations.

In this context, recruiters are being called upon to rethink their mission. Vecchi et al. (2021) argue that it is no longer simply a matter of filling an immediate need, but also of offering a quality candidate experience and implementing a sustainable attractiveness strategy. To do this, they must refocus on the human and relational dimensions of their role, while relying on appropriate technological tools. AI is therefore emerging as an essential lever for transforming the recruitment process. Some companies are already investing heavily in talent acquisition technologies (Agnihotri et al., 2023). This growing use of AI is not limited to a simple modernization of tools, but reflects a broader desire to adapt the HR function to the new demands of the labor market and its candidates.

Recruitment has long been a fundamental pillar of human resources management, playing a central role in attracting what is considered the most valuable resource of any organization: human capital. It is not simply a set of administrative procedures, but a strategic process that determines the quality of internal resources and, consequently, the organization’s overall performance. Numerous studies highlight the importance of viewing recruitment as a competitive lever, emphasizing the need to attract and retain highly qualified candidates to maintain a sustainable market advantage (Hadijah, 2024; Agnihotri et al., 2023; Tulgan, 2015). This process generally begins with human resource planning, which involves a rigorous analysis of the positions to be filled and a precise definition of the skills, abilities, and qualities expected of future employees. Once this phase is completed, successive steps follow, such as candidate sourcing, application screening, selection of the most suitable profiles, and finally, contracting (Brunt, 2016). The effectiveness of these steps determines the quality of hiring decisions, especially since the cost of a failed recruitment can be considerable. Indeed, according to Arms & Bercik (2016), a recruitment error can cost a company up to six times the candidate’s annual salary for a senior position.

Beyond the financial aspect, the organizational consequences of inadequate recruitment are often significant, particularly in terms of team disorganization, demotivation, or loss of productivity. This is why it is becoming essential for companies to perfect their recruitment practices by relying on rigorous, efficient methods and, increasingly, on advanced technologies such as AI, which promise to streamline the process and better match the profiles recruited with the real needs of the positions.

1.2.1. The traditional recruitment process

The traditional recruitment process, which had long been predominant, began to decline in the mid-1990s with the emergence of the internet as a new recruitment tool (Oksanen, 2018). Before this period, the most common method of attracting candidates was to publish advertisements in newspapers, often in sections specifically dedicated to job openings. However, as Otoo et al. (2018) explain, this form of advertising was expensive and priced based on the size of the advertisement, forcing companies to write short and concise messages. Moreover, unlike in the current digital age, job seekers could not simply conduct an internet search to learn more about the company or the position offered. This reality differed sharply from the habits of current generations, particularly millennials, who benefit from instant access to a wealth of online information (Otoo et al., 2018). Previous generations, such as baby boomers, Generation X, and some Generation Y, were accustomed to consulting newspapers and magazines to identify offers and then submitting paper applications (Sollohub, 2019). Sollohub (2019) describes this traditional recruitment as a linear and sequential process, starting with the identification of the desired profile, followed by the attraction of candidates, then the manual sorting of applications, and finally the communication of the results to applicants. This model, although simple, illustrates a recruitment based on non-digital methods, relying mainly on paper exchanges and direct interactions.

1.2.2. The online recruitment process

The advent of the internet has profoundly transformed the recruitment process, providing companies with tools and methods that not only change the way recruiters identify and attract talent, but also the way candidates search for and apply for jobs. This digital revolution has enabled recruiters to rely on a variety of channels, ranging from specialized job sites such as Indeed, Glassdoor, and Monster, to professional social networks such as LinkedIn, not to mention internal company portals. These platforms facilitate the distribution of job advertisements and allow recruiters to be socially active and connected to a vast network of potential candidates (Milovanović et al., 2022). Job seekers, for their part, benefit from simplified access to a wide range of opportunities, which they can consult anytime and from any location, without the need to physically travel. They can thus study the job requirements in detail and discover the benefits offered, which enhances their decision-making (Otoo et al., 2018).

Despite these innovations, the recruitment process retains a familiar structure, with steps similar to those of traditional recruitment. These include defining needs, attracting candidates, screening, interviewing, and making a final decision (Milovanović et al., 2022). The attraction phase, often referred to as sourcing, is crucial and a key driver of recruitment success. Sourcing involves proactively identifying qualified candidates and taking targeted actions to attract them in response to current or anticipated open positions (Atluri & Reddy, 2025). Organizations that consistently invest in this phase see their recruiting capacity significantly improve, outperforming their competitors. Among the preferred platforms for sourcing are LinkedIn, Indeed, and Monster, which allow both posting job ads and directly searching for suitable profiles (Milovanović et al., 2022; Otoo et al., 2018).

Next comes a step often perceived as the most demanding and time-consuming: application screening. This phase consists of a careful analysis of the CVs received to verify the match between candidates’ qualifications and the position requirements. Pre-selection therefore relies on recruiters’ ability to sort through a large amount of information to select only truly relevant profiles (Wheeler & Van de Haterd, 2024; Haga, 2023). To further this assessment, recruiters often conduct an initial telephone contact to gather additional information, particularly on candidates’ motivations and their understanding of the position (Milovanović et al., 2022; Otoo et al., 2018). This step is sometimes enhanced by the implementation of online tests, which can focus on technical knowledge, language skills or specific abilities. These assessments make the selection more objective by directly measuring the candidates’ potential to meet the job requirements (Otoo et al., 2018).

As for interviews, although traditionally they take place in person on company premises, the rise of videoconferencing platforms such as Google Meet or Microsoft Teams has largely democratized remote recruitment, a development that has been particularly accelerated by the health crisis linked to the Covid-19 pandemic (Peterson, 2024). This method offers a significant gain in flexibility, allowing recruiters and candidates to communicate “anytime, anywhere,” without geographical constraints (Chowdhury, 2025). However, according to Wakelin et al. (2024), virtual interviews have certain limitations, notably a quality of interaction and communication that is often reduced compared to face-to-face meetings.

The transmission of emotions and non-verbal signals, essential in the candidate’s assessment, can indeed be altered, which can influence the recruiter’s perception and lead to errors of judgment. Despite these challenges, online interviews are valued for their convenience and ability to speed up the recruitment process. Once the interviews are conducted, the final selection is based on an overall assessment of the candidate’s skills, knowledge, motivation, personality, and interpersonal skills (Wakelin et al., 2024). The goal is to select the person who, beyond technical qualifications, will best fit the company’s culture and environment. This decision is then formalized, most often by email to the successful candidate. It is important to note that even unsuccessful candidates must receive a clear response, in order to adhere to best practices in talent management and employer branding (Otoo et al., 2018).

Thus, online recruitment, while building on the traditional foundations of recruitment, incorporates digital technologies to improve efficiency, reach, and responsiveness. This development clearly illustrates how digital transformation is redefining HR practices, offering new opportunities while posing new challenges to recruitment professionals.

1.2.3. Between traditional recruitment and online recruitment

When we look at the sourcing methods used just a few years ago, we see that they relied primarily on formal and conventional approaches. Companies favored posting advertisements in local or national newspapers, trade journals, or on their own corporate websites (Sjøvaag, 2024). At the same time, other practices were commonly used, such as participating in job fairs, using internal databases, employee recommendations, or previous professional contacts (Boswell et al., 2024; Setúbal et al., 2024). However, these approaches had limited reach, particularly due to their focus on candidates already engaged in a job search process. As the job market has become more complex and competition to attract the best candidates has intensified, these traditional methods have gradually lost their appeal and effectiveness (Boswell et al., 2024; Setúbal et al., 2024). To better understand this evolution, it is useful to refer to the typology proposed by Boswell et al. (2024), which distinguishes three categories of candidates based on their level of engagement in the job search: active candidates, who are actively seeking a new position; semi-passive candidates, who remain attentive to opportunities without being fully engaged; and finally passive candidates, who are not looking to change jobs but may be receptive to an attractive offer. However, traditional methods target almost exclusively active candidates, which considerably reduces the pool of talent available in a competitive environment.

In this context, recruiters have gradually adopted a more dynamic approach, fostered by the rise of digital technologies and social media. The recruitment process, once sequential, now tends to be carried out simultaneously: actions of attraction, sorting and communication with candidates take place in parallel, which accelerates deadlines and increases efficiency (Boswell et al., 2024; Setúbal et al., 2024). Far from completely eclipsing traditional methods, digital tools, and in particular professional social networks, complement them, by offering wider access to passive candidate profiles, increasingly sought after in the contemporary “war for talent” (Sjøvaag, 2024).

1.3. AI in the recruitment process

Although empirical research on AI applied to recruitment remains limited due to the emerging nature of this technology within organizations (Ali & Kallach, 2024; Frai & László, 2021), a growing number of authors already recognize its transformative effects. AI is now seen as an essential lever for automating, accelerating, and optimizing all stages of the recruitment process, from initial sourcing to the final selection of the best candidates (Ali & Kallach, 2024). By relying on advanced big data analysis capabilities, it helps reduce costs, increase decision-making efficiency, and better match skills with job requirements (Allal-Chérif et al., 2021). From the talent identification and attraction phase, AI has quickly established itself as an essential tool, particularly through online recruitment platforms. These platforms, such as LinkedIn, Monster, Indeed, and Job-Builder, use powerful algorithms to match the skills listed in candidate profiles with the specific needs of recruiters (Albassam, 2023).

Thanks to data mining techniques, intelligent engines are able to identify, sometimes proactively, passive candidates whose profiles show strong potential. These systems can not only predict a candidate’s future performance, but also estimate their chances of staying with the company, thus making targeting more precise and strategic (Pan et al., 2023; Allal-Chérif et al., 2021). In addition, they continuously update profiles by automatically collecting data from professional social networks, which strengthens the relevance of recommendations and promotes a more precise match between supply and demand (Chen, 2023). When the process moves to the pre-selection stage, the contribution of AI becomes even more evident. Resume parsing tools and chatbots, these intelligent conversational agents, offer a solution to the time-consuming burden of manual screening (Hunkenschroer & Luetge, 2022). They can initiate exchanges with candidates via email, SMS, or messaging apps to gather key information about their experience, skills, and motivations.

Thanks to advances in natural language processing, these assistants can also answer candidates’ questions about available positions in real time, thus improving their overall experience (Allal-Chérif et al., 2021; Frai & László, 2021). This continuous operation, without time constraints, allows for the simultaneous processing of a large number of applications, freeing up time for human recruiters who can then focus on the most promising profiles (Chen, 2023).

In addition, tools such as Mya, a conversational assistant developed specifically for recruitment, are capable of automating a significant portion of the process, even ranking candidates based on their suitability using AI-based assessment systems (Pan et al., 2023). This type of tool is used by large companies such as L’Oréal and relies on machine learning mechanisms to adjust its analyses over the course of interactions. At the same time, serious games, interactive simulations supported by AI, allow candidates’ behavioral skills to be tested in contexts close to professional reality. They analyze, in particular, stress management, the ability to make decisions, or innovate, while circumventing biases related to appearance, qualifications, or age (Allal-Chérif et al., 2021; Hunkenschroer & Luetge, 2022). During the interview process, automated video interviewing technologies, such as those offered by HireVue, take AI integration even further (Ajunwa, 2021). These systems analyze candidates’ verbal and nonverbal language, posture, facial expressions, and tone variations, to assess their compatibility with the position and company culture (Albassam, 2023). Some tools even detect signs of deception through micro-eye movements, thus strengthening the behavioral assessment process.

Finally, in the decision-making phase, AI algorithms facilitate the final selection process by ranking candidates based on objective criteria. This approach reduces cognitive biases that can cloud human judgment, such as personal affinity or stereotypes, and promotes an assessment based solely on professional aptitude. By neutralizing discriminatory factors such as origin, age, or gender, AI thus offers the promise of fairness in recruitment, while contributing to greater operational efficiency (Hunkenschroer & Luetge, 2022; Frai & László, 2021).

Although AI is often presented as a tool for neutralizing human bias in recruitment processes, several authors highlight the limitations of such an optimistic vision. Indeed, if AI relies on pre-existing data, this data can incorporate historical distortions or discriminatory patterns, thus leading to a systemic reproduction of the biases it was intended to correct (Hunkenschroer & Luetge, 2022; Frai & László, 2021). The emblematic case of Amazon, which developed an algorithm based on ten years of internal data, shows how a system trained primarily on male applicants ended up systematically discriminating against women (Ajunwa, 2021). This situation reveals a fundamental truth: algorithms are neither objective nor independent of their designers, but reflect the inherent biases of the data on which they are based.

Furthermore, the widespread use of AI in recruitment raises fears of a cloning effect. By relying on correlations deemed statistically relevant, algorithms favor the selection of similar profiles, which can lead to a homogenization of teams, to the detriment of diversity and innovation. Atypical, unconventional but potentially innovative profiles then risk being systematically rejected (Albassam, 2023; Allal-Chérif et al., 2021). This formatting phenomenon raises critical issues regarding organizations’ ability to identify emerging talent, integrate new perspectives, and evolve in a constantly changing world.

Beyond these limitations, the adoption of AI in HR practices faces technological, cultural, and ethical challenges. The complexity of the tools, the need for advanced digital skills, and the fear of loss of control or intrusive surveillance are generating resistance among both decision-makers and employees (Pan et al., 2022; Hunkenschroer & Luetge, 2022). Automatically extracting information from social networks to predict a candidate’s behavioral compatibility, while effective, raises fundamental questions about privacy, transparency, and the ethics of algorithmic processing (Hunkenschroer & Luetge, 2022).

RESEARCH METHODOLOGY

As part of this research, which aimed to explore Tunisian recruiters’ perceptions of the increasing integration of AI in the recruitment process, it seemed essential to adopt a qualitative approach. This methodology emerged as the most relevant as it allows for an in-depth understanding of the representations, discourses, and practices of the stakeholders involved, beyond simple quantitative data. This approach is fully in line with a comprehensive perspective, seeking to capture the experiences of professionals in their specific context and to account for how they approach these technological transformations (Shahruddin & Husain, 2024).

In accordance with the methodological recommendations put forward by Tracy (2024), qualitative research is carried out in several successive phases: formulation of the problem, collection of empirical material, data analysis, and then interpretation and reporting of the results. Having previously defined our research question, the next phase consisted of developing a data collection strategy that would allow us to compare our hypotheses with realities on the ground.

To do this, we opted for semi-structured interviews, a method particularly suited to the exploratory objectives of this study. This choice was based on the desire to offer interviewees the opportunity to express themselves freely on the topics discussed, while maintaining a framework to ensure comparability. This flexibility not only allows us to follow the thread of the discussion, but also to capture unexpected elements and explore certain points based on the responses received, thus making the discussions richer and more nuanced (Lim, 2025).

The survey was conducted with ten professionals working in Tunisian companies operating in various sectors such as services, information technology, manufacturing, and distribution. This choice of sector diversity aims to capture a broad range of perceptions and experiences, while ensuring a certain methodological consistency to the extent that these companies share similar challenges in terms of recruitment and digital transformation. The sample consists mainly of human resources professionals directly involved in recruitment, such as HR managers, talent acquisition specialists and HR business partners. To enrich our technical understanding of the use of AI, we also interviewed two data scientists as well as an automation expert, which allowed us to cross-reference the perspectives of HR specialists and technology experts.

Participants were selected through our professional network in Tunisia, particularly via LinkedIn, as well as through the so-called “snowball” method, which involves soliciting recommendations from those already interviewed. Out of a total of 25 people contacted, ten agreed to participate in this study. To maintain the confidentiality of their comments, participants were identified by their position and a number corresponding to the order in which the interviews were conducted. The profiles of these respondents varied in terms of age, years of experience, and level of familiarity with digital tools, which contributes to the richness of the data collected.

The majority of interviews were conducted remotely, while four were conducted in person, which allowed us to adapt to the participants’ availability and overcome geographical constraints. Each interview began with a presentation of the research objectives and an explicit request for consent to record the discussion, with anonymity and confidentiality guaranteed. This preliminary step fostered a climate of trust conducive to open and authentic exchanges.

Table 1. Characteristics of the Qualitative Sample and Data Collection Process

The topics covered during the interviews were organized around several axes: the description of the participant’s role, the types of profiles recruited, the challenges encountered in the Tunisian job market, the various stages of the recruitment process, and perceptions related to the introduction of AI. Discussions also focused on the expected benefits, potential risks, and the ethical and organizational challenges associated with this transformation. This thematic framework, although structured, evolved over the course of the interviews, with certain elements being explored in more depth based on the respondents’ contributions.

Once the data collection phase was completed, all interviews were transcribed in full. As Tracy (2024) points out, transcription is a crucial step in the process, as it preserves the richness and fidelity of the discussions while facilitating rigorous and systematic analysis. This requirement for rigor aims to avoid any loss of potentially relevant information, even when certain data may seem secondary at first glance.

Data analysis was conducted using an inductive thematic approach, inspired by Tracy’s (2024) method. This process took place in several stages: first, open coding identified emerging meaning units in participants’ discourses. These initial codes were then grouped into broader categories, leading to the construction of cross-cutting themes reflecting the main concerns expressed. These themes include the perception of AI as a lever for optimization or, conversely, as a source of fear for the future of HR professions, the shift in skills expected of Tunisian HR professionals, the transformation of recruitment practices, as well as ethical questions and dilemmas related to the collection and processing of personal data. This entire process has made it possible to produce a detailed and nuanced vision of the dynamics at work in the adoption of AI within HR functions in Tunisia, while revealing the tensions, hopes and resistance that this technological transition arouses in the daily lives of recruiters.

RESULTS AND DISCUSSION

3.1 Recruitment

The analysis of interviews conducted with ten Tunisian professionals specializing in recruitment highlights a shared reality: the job market is currently particularly tense. This tension results from both a talent shortage and the increasing bargaining power of candidates (Haga, 2023). The dynamic has reversed, now placing companies in the position of having to attract candidates, and not the other way around.

As one recruiter points out, candidates today have a wide range of opportunities, which makes them particularly demanding:

“We realize that today, candidates have many offers at the same time, and therefore we have to be able to sell our client to be sure they will go with them. They have much more choice than before and greater bargaining power. They allow themselves to make demands that are too high for their expertise, particularly in terms of salary.” This poses challenges for companies that must adapt to this new context.” (Interview 3)

A recruiter in the financial sector expressed a similar concern, highlighting the lack of qualified profiles:

“A candidate can have up to five offers at the same time. So, they have a choice, and it’s up to the organizations to sell themselves. Finding profiles, whether junior or experienced, is extremely difficult. It’s not because people don’t want to work, but because there aren’t enough people with the right skills. The market is tight. The candidate has the power, the choice, and the time to choose where they want to go.” (Interview 7)

This phenomenon is accentuated by a change in candidate mindsets, who are placing increasing emphasis on organizational culture and company values (Otoo et al., 2018). As one recruiter summarizes:

“Today, the market is tight, and candidates are more sophisticated in their requests. They want to ensure that the company shares their values and culture. These expectations are becoming increasingly common. If candidates aren’t interested in the industry or what the company offers, it’s very difficult to attract them. They know what they want.” (Interview 5)

This shift in expectations isn’t limited to compensation. It now includes the search for a better work-life balance, flexibility, and suitable working conditions (Fasang & Aisenbrey, 2022). Another professional explains:

“People are looking more for a balance between their personal and professional lives, even young people, and I think that’s healthy. Afterwards, you have to find the right balance. With the health crisis, people, regardless of their age, are asking questions: what is your policy on teleworking, flexibility, etc.?” (Interview 5)

One of the most experienced recruiters also testified to a profound generational shift in the relationship with work:

“I’ve been working for 15 years, and there’s clearly been a shift in mentality. Today, especially among younger people, the question is: what are you going to give me, and what can I expect from you? The job for life is over. They take what they need from the company, and if it doesn’t suit them, they move on.” (Interview 5)

While the steps in the recruitment process haven’t fundamentally changed—posting the job offer, screening applications, and interviews—their duration has increased, particularly due to the difficulty of finding certain profiles. Several recruiters explain that the number of interviews has increased; sometimes three to four steps are required, especially for specific or technical profiles. The increasing digitalization of processes and the gradual introduction of AI, for their part, are generating mixed opinions (Kodieswari et al., 2024). On the one hand, these technologies allow for the automation of certain tasks and increased efficiency. On the other, they contribute to a dehumanization of the process, as one professional laments:

“A distance has been created with the candidate. Before, people would submit their CVs directly in person. Today, everything goes through a computerized process. We respond to online advertisements, and now with AI, there are automatic selections that decide whether the candidate is suitable or not. We are moving further and further away from human contact.” (Interview 3)

This observation is reinforced by the difficulties encountered with the widespread use of videoconference interviews since the pandemic. Several professionals point out that this tool, while practical, does not allow for a full understanding of candidates’ nonverbal language and emotions (Peterson, 2024). One recruiter explains:

“During Covid, it wasn’t possible to do otherwise, but seeing the candidate in person changes everything. You can see if they’re nervous, if they’re moving their legs, if they’re breathing faster. These are signs you don’t get in a video interview. Plus, it helps build rapport and give them confidence. These are things that come across much better in person.” (Interview 3)

Another recruiter agrees, pointing out the technical limitations:

“Just doing a virtual interview or a face-to-face interview is not the same at all. In a video interview, you only see the person’s face, it’s harder to get a sense of what’s going on. And then there are camera cutouts, sound issues… it’s very disruptive.” » (Interview 1)

However, despite these limitations, several professionals recognize that videoconferencing offers undeniable advantages in terms of flexibility. One recruiter confides:

“I was uncomfortable with my first five video interviews during Covid, but today I think it’s great. The candidate can connect whenever they want, and so can I, from wherever I want. It’s much simpler.” (Interview 2)

Another adds:

“It’s true that sometimes we have candidates who work until 6 p.m. Seeing them in person afterward is complicated. With a Teams interview, it’s easier for everyone. We adapt.” (Interview 3)

Nevertheless, for the majority of recruiters surveyed, it remains essential to schedule an in-person interview, at least at a late stage in the process. This return to face-to-face contact is seen as essential to truly assess the compatibility between the candidate and the company, both in terms of skills and values. As one professional summarizes:

“We used video interviews during Covid, but now we’re going back to face-to-face interviews. I think the candidate should at least come to the company once, meet the people, and see if it’s going well in person. This allows the candidate to get a feel for the environment and allows us to better understand the person. Face-to-face, the conversation is more natural, and we detect other things.” (Interview 10)

Indeed, the testimonies collected reveal that the recruitment market in Tunisia is undergoing rapid change. Between growing candidate expectations, the digitalization of processes, and the challenges related to skills shortages, HR professionals must constantly adapt their practices to remain competitive and attractive.

AI

When we asked Tunisian professionals about their understanding of AI, their responses converged around the same general definition. For them, AI is essentially perceived as an algorithm capable of learning, a decision-making tool based on data mining, and which attempts to reproduce certain mechanisms of human intelligence. One of the interviewees summarized this perception by explaining that AI consists of:

“Exploiting data in such a way as to automate certain tasks, while approaching, to a certain extent, our own way of thinking. These are algorithms that react almost like a human would.” (Interview 8)

However, it appears that, in the Tunisian context, the human resources professionals we met do not currently use HR software that fully integrates AI. Moreover, they consider these technologies to be more accessible to large multinational companies than to local organizations (Daugherty & Wilson, 2024). However, some acknowledge that platforms they use daily, such as LinkedIn, already integrate AI-based features, even if this often goes unnoticed. One recruiter shares:

“Even if I don’t directly use an AI tool, when I create a project on LinkedIn, the platform suggests profiles I wouldn’t necessarily have found on my own. These are often very relevant suggestions, so for me, that’s already AI.” (Interview 7)

The question of using more advanced tools, such as chatbots or video interview systems with facial expression analysis, is divisive (Ghosh et al., 2024; Allal-Chérif et al., 2021). Many believe it would be beneficial to have this type of assistance, particularly to detect certain weak signals that can go unnoticed during a traditional interview. However, they insist that the final decision cannot be left to a machine. One recruiter expresses this idea as follows:

“AI can help us pick up certain micro-signals, but it can’t replace human contact. We have to have the final say, because if the candidate never sees anyone and only talks to machines, it doesn’t reflect well on the company.” (Interview 9)

The majority believe that AI could be useful primarily for time-consuming administrative tasks, such as sorting applications, writing reports, or verifying basic criteria (Hunkenschroer & Luetge, 2022). One professional illustrates this idea with a concrete example:

“I had a position where fluency in Spanish was mandatory, and yet I received several CVs from people who didn’t speak Spanish. An AI could filter this automatically and save me valuable time.” Two minutes here, two minutes there… in the end, it’s time we could use for more important tasks.” (Interview 1)

When discussing the possibility that AI could limit certain cognitive biases present among recruiters, several participants indeed believe it could play a positive role by ensuring a certain neutrality. A recruiter shared this reflection:

“I remember a colleague who often favored candidates from his own village or with similar hobbies. We are all, consciously or not, influenced by personal details. AI could be a solution to reduce this type of bias, provided it is properly configured.” (Interview 5)

The limitations of AI in recruitment are also repeatedly mentioned. Several professionals note that an algorithm could reject interesting profiles simply because something is missing from the CV (Ajunwa, 2021). One of them illustrates this:

“A person may have forgotten to mention that they speak English, but looking at their background and the companies they have worked for, you can guess that they are fluent in the language.” AI, on the other hand, would rule it out directly.” (Interview 3)

More generally, some express the fear that AI will tend to homogenize profiles, by systematically favoring the same criteria and leading to a form of standardization of candidates. As one recruiter indicates:

“I’m a little afraid that AI will offer us clones, profiles that are too bland, too uniform. I don’t know if that will happen, but it’s a risk I can imagine.” (Interview 10)

Another widely shared point concerns the difficulty for AI to take into account interpersonal aspects and cultural compatibility, particularly in small teams where personality plays a key role. One recruiter explains:

“In a team of five people, personality is essential. If someone doesn’t share the same mindset, they won’t be able to integrate. And AI will never see that.” » (Interview 3)

Finally, several participants pointed out the practical obstacles associated with implementing these tools in Tunisia. They notably mentioned the high cost, training requirements, and adapting mindsets. One recruiter emphasized:

“If we want to integrate a system like this, it has to be simple and intuitive, because not everyone is necessarily comfortable with technology, especially colleagues with many years of experience. We’ll have to allocate time to train them.” (Interview 4)

The issue of data protection also emerges as a major concern. For many, the use of AI in recruitment raises important ethical questions, particularly regarding the management of personal information. One recruiter emphasized:

“If it’s not well regulated, it could pose major problems, both for companies and candidates. Data protection will become a crucial issue in the coming years.” » (Interview 6)

Thus, the feedback from these ten Tunisian professionals reveals a nuanced relationship with AI: perceived as a lever for optimizing certain technical aspects of recruitment, it nevertheless remains largely insufficient to replace human judgment, understanding of contexts, and emotional intelligence, considered fundamental elements of the recruitment process (Oksanen, 2018).

3.2 Recruitment and AI

When asked whether AI could eventually replace the recruiting profession, all the Tunisian professionals surveyed agreed that this possibility remains unlikely (Chowdhury, 2025). For them, the heart of recruitment lies in human interaction, exchange, listening, and intuition, dimensions that machines cannot replicate. They recognize, however, that AI will play an increasingly important role in optimizing repetitive and administrative tasks, such as sorting CVs or pre-selecting candidates. However, they emphasize that this automation will be limited to the early stages of the process, leaving humans to take charge of the decisive phases where interpersonal skills, empathy, and affinity assessment play a central role (Albassam, 2023). “I think that’s already the case for some professions. I mean, for example, there are fewer and fewer salespeople at the checkout in stores because we can do it ourselves, or there are fewer and fewer people packing boxes because machines do it. So, I think there are professions for which there will no longer be a need for humans, but on the other hand, precisely in recruitment, from the moment we touch on relationships, we need a human.” (Interview 10)

Another professional compares AI to a support system that assists the recruiter in their decision-making, without ever replacing them.

“I think humans still have their place. AI isn’t designed to replace humans; it’s really there to help us, to support us, I mean. It’s something that will assist us, like a colleague. It will really help us make the right decisions.” It will intervene in the early stages, precisely to assist in decision-making, but the final decision will always be human. Humans must always be at the heart of the profession.” (Interview 9)

This conviction is also shared by another recruiter, who emphasizes the essential role of feelings and intuition in the recruitment process.

“For me, AI only comes into play at the beginning of the process to select the first candidates. Then, in the interview, the person’s technical skills are taken into account, but personality also plays a major role, and I find that only human beings can sense whether it will work or not.” (Interview 4)

Analysis of the comments collected from human resources professionals in Tunisia shows that they have a fairly good understanding of AI, despite the inherent complexity of the subject. For most, AI is perceived as a set of algorithms capable of exploiting data in order to automate certain tasks and simulate forms of human intelligence (Ali & Kallach, 2024). However, when it comes to its concrete application in their daily practices, particularly in recruitment, the majority say they do not directly use tools specifically designed on the basis of AI (Frai & László, 2021).

A particularly interesting point emerges from the interviews: many of them use platforms such as LinkedIn daily, without necessarily realizing that this tool already incorporates AI-based features (Milovanović et al., 2022; Otoo et al., 2018). This lack of awareness raises a first limitation regarding the actual perception of AI in the Tunisian professional context.

“For me, AI is everything related to data mining, which enables automation and is closer to our way of thinking. It’s intelligent programming that aims to create systems capable of reproducing, to a certain extent, our own reasoning.” » (Interview 8)

While some acknowledge that LinkedIn offers profile suggestions based on machine learning algorithms, none of the professionals interviewed mentioned using more advanced tools, such as pre-screening chatbots or video interviews with behavioral analysis, although these tools are documented in international literature (Ali & Kallach, 2024; Allal-Chérif et al., 2021).

“I think all recruiters use LinkedIn, and when I create a recruitment project, the tool regularly suggests profiles I hadn’t identified. For me, this is already a form of AI.” (Interview 7)

Overall, our interviewees expressed strong expectations for AI to reduce the burden of administrative tasks: scheduling interviews, processing applications, sorting resumes, and even managing emails. They consider these tasks time-consuming and create little added value, which, according to them, fully justifies the integration of AI in these specific areas.

“AI would save us a lot of time. I’m thinking in particular of verifying basic criteria, such as the languages ​​required for a position. Too often, we receive CVs that don’t meet basic criteria.” (Interview 1)

Nevertheless, as soon as the question of replacing human assessment arises, particularly during interviews where personality and feelings play a key role, professionals are unanimously cautious (Vassallo et al., 2024). They believe that AI cannot capture the subtleties of human relationships, nor detect the weak signals that allow them to assess a candidate’s compatibility with a team.

“AI could detect certain micro-details that we, as recruiters, may miss, but it will never replace human contact.” The feeling, the connection that can be created with a candidate, that’s something no machine can do.” (Interview 9)

The issue of bias is also central to their thinking. While some acknowledge that AI could limit unconscious bias (Priyadarshini & Cotton, 2021), others believe that these same biases are an integral part of human nature and the recruiting profession.

“We can unconsciously favor a candidate who resembles us. AI could neutralize that, but the data it relies on must itself be neutral.” (Interview 5)

Others, on the contrary, see this neutrality as a form of coldness, likely to impoverish the recruitment process by removing essential subjective elements, such as feeling or intuition.

In addition, several major limitations are highlighted. First, AI’s inability to detect atypical profiles that might otherwise meet the job requirements despite an incomplete CV or one that doesn’t meet predefined criteria (Wheeler & Van De Haterd, 2024).

“If someone forgets to specify that they speak English on their CV, we, looking at their experience in certain companies, can guess and call them. AI would have automatically eliminated them.” (Interview 3)

Second, the risk of homogenizing recruited profiles frequently comes up in discussions. Recruiters fear that excessive use of AI will lead to the creation of overly uniform teams.

“I’m afraid that AI will end up providing us with clones, profiles that are too bland, too similar.” (Interview 10)

Furthermore, the cost of these technologies remains a significant obstacle for Tunisian companies, particularly SMEs (Arms & Bercik, 2016). Our interviewees emphasize that the profitability of such an investment depends directly on the volume of recruitment carried out. In other words, the value of AI is less if the company only recruits occasionally (Boswell et al., 2024).

“The robot needs to be programmed, and it’s only worthwhile if you recruit regularly for the same types of positions. Otherwise, the investment doesn’t really make sense.” (Interview 8)

Finally, the question of the acceptability of these tools is closely linked to employee training and companies’ ability to develop skills. Our interviewees believe that this change must be supported while taking into account resistance, without age necessarily being a determining factor.

Ultimately, our results show that Tunisian professionals are still cautious about integrating AI into recruitment. While they recognize its potential to automate certain tasks, they remain convinced that human judgment, personal experience, and the quality of the relationship remain irreplaceable.

Table 2: Summary of Key Themes on recruitment and AI from Interviews (N = 10)

Key themes and trends from the Tunisian findings could also be used to benchmark against other emerging economies in terms of assessing adoption barriers and perceptions of AI in recruitment.

3.3 Adoption Barriers and Perceptions

Job Market Friction — As in many developing economies, Tunisia is under considerable job market stress. This is due largely to high unemployment rates — especially for youth — alongside an influx of fresh graduate.

Evolving Talent Landscape: While the demands for better job conditions and flexibility, along with the choice of work from home, ability to travel or visibility during any career transition still continues, other emerging markets, too, share this challenge of balancing out the employer and candidate expectations, which is another factor leading towards the competition for talent acquisition.

Qualified Profiles: One of the emerging economies worldwide always have the pain of struggling with identifying qualified candidates. Many governments will also grapple with the challenges of the skills gap and inequality in education systems. Fifth in Tunisia, where 80% of candidates facing difficulty to find qualified profiles.

Length of Recruitment Process: As has also been noted in other areas, the length of the recruitment process has extended in some cases and shortened in others as a result of what is happening in the market. One common purpose of digitalization is to shrink these timespans, but the pace of adoption can differ.

Digitalization & Videoconferencing: More and more people are combining videoconferencing into their digital toolbox. Tunisia has widely adopted those tools (80% of mention), however other emerging economies also report such a change (faster due to COVID-19 pandemic).

Familiarity with AI and Usage of AI: The general knowledge around AI and its use in recruitment has improved marginally but still remains inconsistent. While all Tunisian interviewees (100%) show familiarity with AI in recruitment usage, the scope and applicability of AI can be quite disparate from one country to another.

AI Replacement of Human Judgment: A fear for general AI replacement of human judgment (100% in Tunisia) confirms anxiety in other emerging economies, especially about bias and confidence in automated decisions.

3.4 AI-Enabled Recruitment Solutions in Tunisia

Like other emerging economies, multiple AI-enabled recruitment solutions are making strides in Tunisia:

LinkedIn Recruiter — One of the most popular sourcing platforms. Recruiters can sift through millions of candidates, filter for qualification or skill set, and get AI-based suggestions to zero-in on relevant profiles. On LinkedIn, every job post gets a higher visibility with prompts to recruiters to connect with the candidates.

AI Screening in Applicant Tracking Systems (ATS) — Organizations are now leveraging an ATS that uses AI to screen resumes. Using data-powered technology, these systems will compare job descriptions to candidate applications and highlight relevant skills and qualifications in applications that are a close match to the job, then filter those that don’t meet the hiring criteria. This aids in making the first steps of recruitment more efficient.

As recruiters move towards chatbots to assist candidates in real time who are seeking information about the job posting Chatbots: Increasingly Tunisia has started finding Chatbots are being used to assist in real time for candidates seeking information w.r.t jobs. By quickly responding to commonly asked questions and scheduling interviews, these AI-powered tools facilitate increased engagement between candidates and HR, leading to a faster hiring process.

3.5 Contextual Insights

Although the use of AI recruitment tools in Tunisia has great potential, the following contextual elements influence their effectiveness:

Digital literacy: The state of literacy, both on the employer and the job seeker side, will really determine how well the tools are adopted and optimized. This does not mean that the regions with lower literacy rates will be slower in adoption.

— Infrastructure — The level of internet access and the technological infrastructure level that your area will impact the utility capabilities of each of these tools, particularly in rural locations and where connectively may be limited.

Cultural Outlooks: The general perception about technology and automation varies a lot, so there can be a lot of differential trust and acceptance of AI tools on hiring processes compare to other countries.

Overall, while some of the challenges and opportunities that other developing nations experience with applications of AI in recruitment are relevant for Tunisia, the specifics of the national context, including cultural perceptions and digital readiness, are key in determining the process and success of AI diffusion through HRM.

CONCLUSION

This qualitative research, conducted with ten human resources professionals working in Tunisia, provided a better understanding of the perceptions, actual uses, and representations associated with AI in the recruitment field. Through an inductive approach based on semi-structured interviews, our objective was to give practitioners a voice in order to understand their relationship with this technology, which is still emerging in the Tunisian context. This qualitative methodology, based on a thematic analysis of discourses, offered us the opportunity to explore in depth not only conscious perceptions, but also the unspoken issues, hesitations, and ambivalences that accompany reflections on AI applied to human resources.

One of the first major lessons of this study lies in the observation that, despite the presence of generally informed discourse on the concept of AI, its adoption in recruitment processes remains very limited in Tunisia. The majority of professionals surveyed admitted to using digital tools such as LinkedIn, without being aware that these platforms already integrate features powered by AI algorithms (Milovanović et al., 2022). This partial lack of understanding of AI, combined with a certain confusion between simple automation and algorithmic intelligence, reflects a low level of technological maturity in the Tunisian HR sector.

 In the Tunisian context, the reality appears more nuanced, marked by practices that are still largely traditional. In terms of contributions, this study highlights several crucial points. First, it underlines that Tunisian professionals do not perceive AI as a threat to their role, but rather as a support tool, essentially intended to automate repetitive and time-consuming tasks such as sorting CVs, managing applications or scheduling interviews (Suhonen, 2025; Benabou et al., 2024; Kwon & Jang, 2022). AI is therefore seen as a lever for operational efficiency, without encroaching on the human dimension of recruitment, which is considered irreplaceable. This result confirms that the heart of the recruiter’s job, which is based on creating a human connection, understanding behavioral subtleties and assessing a candidate’s cultural compatibility, cannot be entrusted to an algorithm, at least in the short term.

Furthermore, the study reveals shared concerns regarding algorithmic bias, the risks of profile standardization, and the difficulty of AI in handling atypical cases or soft skills, which are nevertheless crucial to recruitment success (Díaz-Rodríguez et al., 2023). Our interviewees also expressed ethical concerns, particularly regarding the protection of personal data, the transparency of algorithms, and compliance with current regulations, which are still poorly structured in Tunisia.

On the economic front, a major obstacle remains the cost of acquiring and deploying AI tools, often considered disproportionate to the volume of recruitment carried out by the majority of Tunisian companies, particularly SMEs (Arms & Bercik, 2016). This economic dimension, little explored in the scientific literature, appears to be a significant obstacle in our context. This observation leads to a relevant reflection on the relationship between recruitment volume and the perceived usefulness of AI: the more applications a company processes, the more relevant and profitable the benefits of an intelligent tool become. This connection between volume and added value constitutes, in our opinion, a unique contribution of this research, which could inform theoretical models on technology adoption in HR.

Nevertheless, this study has several limitations that should be carefully noted. First, the relatively small size of our sample, composed of ten professionals, while adequate for exploratory qualitative research, does not allow us to generalize the results to the entire Tunisian economic fabric. Furthermore, the diversity of the sectors represented remains partial, which could influence certain perceptions related to sectoral specificities. Furthermore, the study focused exclusively on the perspective of recruiters without incorporating that of candidates, who could have provided additional insight into the reception and perceived legitimacy of AI in the recruitment process. Methodologically, the qualitative approach, while providing a rich description and a detailed understanding of the phenomena, does not provide quantitative data to measure the extent of the observed trends. Continuing this research using a mixed methodology, combining quantitative surveys with a larger sample, would strengthen the robustness of the results and refine the analyses.

Ultimately, this research contributes to the literature by offering a perspective anchored in the Tunisian context, where AI in HR remains a work in progress. It serves as a reminder that the adoption of AI in recruitment is not simply a technological issue, but also involves strategic, ethical, economic, and human choices. The results show that the acceptability of these tools will largely depend on their ability to complement human skills rather than replace them. For researchers, this study opens up promising avenues, particularly around the formalization of an AI acceptance model adapted to the specificities of emerging countries like Tunisia, integrating rarely explored variables such as recruitment volume or the degree of digital maturity of companies. For practitioners, it highlights the urgency of raising awareness, training, and supporting HR teams to better understand AI technologies, while ensuring that the fundamentals of the profession are preserved: human, relationships, and discernment.

BIBLIOGRAPHIC REFERENCES

1. Agnihotri, A., Pavitra, K. H., Balusamy, B., Maurya, A., & Bibhakar, P. (2023). Artificial intelligence shaping talent intelligence and talent acquisition for smart employee management. EAI Endorsed Transactions on Internet of Things, 10.
2. Ajunwa, I. (2021). Automated video interviewing as the new phrenology. Berkeley Technology Law Journal, 36(3), 1173-1226.
3. Albassam, W. A. (2023). The power of artificial intelligence in recruitment: An analytical review of current AI-based recruitment strategies. International Journal of Professional Business Review: Int. J. Prof. Bus. Rev., 8(6), 4.
4. Ali, O., & Kallach, L. (2024). Artificial Intelligence Enabled Human Resources Recruitment Functionalities: A Scoping Review. Procedia Computer Science, 232, 3268-3277.
5. Allal-Chérif, O., Aránega, A. Y., & Sánchez, R. C. (2021). Intelligent recruitment: How to identify, select, and retain talents from around the world using artificial intelligence. Technological Forecasting and Social Change, 169, 120822.
6. Arms, D., & Bercik, T. (2016). 10 hiring mistakes to avoid. Strategic Finance, 97(7), 46.
7. Atluri, A., & Reddy, V. (2025). Cognitive HR Management: How Oracle HCM Is Reinventing Talent Acquisition through AI. International Journal of Artificial Intelligence, Data Science, and Machine Learning, 6(1), 85-94.
8. Augusto, L. M. (2021). From symbols to knowledge systems: A. Newell and HA Simon’s contribution to symbolic AI.
9. Bardozzo, F., Terlizzi, A., Simoncini, C., Lió, P., & Tagliaferri, R. (2024). Elegans-AI: How the connectome of a living organism could model artificial neural networks. Neurocomputing, 584, 127598.
10. Bellaby, R. (2024). The ethical problems of ‘intelligence–AI’. International Affairs, 100(6), 2525-2542.
11. Benabou, A., Touhami, F., & Demraoui, L. (2024, May). Artificial intelligence and the future of human resource management. In 2024 International Conference on Intelligent Systems and Computer Vision (ISCV) (pp. 1-8). IEEE.
12. Bidyalakshmi, T., Jyoti, B., Mansuri, S. M., Srivastava, A., Mohapatra, D., Kalnar, Y. B., … & Indore, N. (2024). Application of Artificial Intelligence in Food Processing: Current Status and Future Prospects. Food Engineering Reviews, 1-28.
13. Bory, P., Natale, S., & Katzenbach, C. (2024). Strong and weak AI narratives: an analytical framework. AI & SOCIETY, 1-11.
14. Boswell, W. R., Payne, S. C., & Bowman-Callaway, C. E. (2024). Recruiting employed job candidates. In Essentials of Employee Recruitment (pp. 171-193). Routledge.
15. Brunt, C. (2016). Human resource management in International NGOs. New York, NY: Palgrave Macmillan.
16. Chen, Z. (2023). Collaboration among recruiters and artificial intelligence: removing human prejudices in employment. Cognition, Technology & Work, 25(1), 135-149.
17. Chowdhury, R. H. (2025). Digital leadership and organizational learning: Technologies for business transformation and operational excellence. Deep Science Publishing.
18. Dandotiya, A. S., Gupta, S. K., Dandotiya, N., & Sharma, M. P. (2024). AI IN EVERYDAY LIFE: TRANSFORMING SOCIETY. Navi International Book Publication house.
19. Daugherty, P. R., & Wilson, H. J. (2024). Human+ Machine, Updated and Expanded: Reimagining Work in the Age of AI. Harvard Business Press.
20. Díaz-Rodríguez, N., Del Ser, J., Coeckelbergh, M., de Prado, M. L., Herrera-Viedma, E., & Herrera, F. (2023). Connecting the dots in trustworthy Artificial Intelligence: From AI principles, ethics, and key requirements to responsible AI systems and regulation. Information Fusion, 99, 101896.
21. Dobbin, F., & Kalev, A. (2022). Getting to diversity: What works and what doesn’t. Harvard University Press.
22. Elkholy, E. M., Abo Gad, R. A. E., & Mahmoud, S. A. (2025). Nurse Managers’ Perception Regarding Artificial Intelligence and Health Logistic Management. Tanta Scientific Nursing Journal, 36(1).
23. Fasang, A. E., & Aisenbrey, S. (2022). Uncovering social stratification: Intersectional inequalities in work and family life courses by gender and race. Social Forces, 101(2), 575-605.
24. Frai, J., & László, V. (2021). A literature review: artificial intelligence impact on the recruitment process. International Journal of Engineering and Management Sciences, 6(1), 108-119.
25. Ghosh, S., Ness, S., & Salunkhe, S. (2024). The role of AI enabled chatbots in omnichannel customer service. Journal of Engineering Research and Reports, 26(6), 327-345.
26. Gjølstad, J., & Skants, N. A. S. (2023). Artificial Intelligence in Organizational Decision-Making (Master’s thesis, Handelshøyskolen BI).
27. Gonçalves, B. (2023). The Turing test argument. Taylor & Francis.
28. Gupta, R. (2024). Impact of Artificial Intelligence (AI) on Human Resource Management (HRM). International Journal For Multidisciplinary Research, doi, 10.
29. Hadijah, H. S. (2024). Implementation of talent management as a strategy for achieving company competitive advantage. International Journal of Artificial Intelligence Research, 6(1.1).
30. Haga, V. (2023). Recruiter perception of bias in screening processes and ways to avoid it (Master’s thesis, Hanken School of Economics).
31. Hetrick, R., Grieser, H., Sentz, R., Coffey, C., Burrow, G., & Botkin, D. (2021). The demographic drought. Emsi. https://www. economicmodeling. com/wp-content/uploads/2021/07/Demographic-Drought-V18. pdf.
32. Horowitz, M. C., Allen, G. C., Kania, E. B., & Scharre, P. (2022). Strategic competition in an era of artificial intelligence. Center for a New American Security..
33. Huang, X., Yang, F., Zheng, J., Feng, C., & Zhang, L. (2023). Personalized human resource management via HR analytics and artificial intelligence: Theory and implications. Asia Pacific Management Review, 28(4), 598-610.
34. Hunkenschroer, A. L., & Luetge, C. (2022). Ethics of AI-enabled recruiting and selection: A review and research agenda. Journal of Business Ethics, 178(4), 977-1007.
35. Jarrahi, M. H. (2018). Artificial intelligence and the future of work: Human-AI symbiosis in organizational decision making. Business horizons, 61(4), 577-586.
36. Jha, N., Prashar, D., & Nagpal, A. (2021). Combining artificial intelligence with robotic process automation—an intelligent automation approach. Deep learning and big data for intelligent transportation: enabling technologies and future trends, 245-264.
37. Johnson, D. G., & Wetmore, J. M. (Eds.). (2021). Technology and society: Building our sociotechnical future. MIT press.
38. Kodieswari, A., Sabarmathi, K. R., Remya, K., Gayathiri, N. R., Nithyapriya, S., & Malathi, T. (2024). Statistical AI Model in an Intelligent Transportation System. In Artificial Intelligence for Future Intelligent Transportation (pp. 145-171). Apple Academic Press.
39. Korteling, J. H., van de Boer-Visschedijk, G. C., Blankendaal, R. A., Boonekamp, R. C., & Eikelboom, A. R. (2021). Human-versus artificial intelligence. Frontiers in artificial intelligence, 4, 622364.
40. Kuipers, M., & Prasad, R. (2022). Journey of artificial intelligence. Wireless Personal Communications, 1-16.
41. Kvint, V. L., & Bodrunov, S. D. (2022). Strategizing societal transformation: Knowledge, technologies, and noonomy. Apple Academic Press.
42. Kwon, K., & Jang, S. (2022). There is no good war for talent: A critical review of the literature on talent management. Employee Relations: The International Journal, 44(1), 94-120.
43. Li, X., Shi, H., Xu, R., & Xu, W. (2025). Ai awareness. arXiv preprint arXiv:2504.20084.
44. Lim, W. M. (2025). What is qualitative research? An overview and guidelines. Australasian Marketing Journal, 33(2), 199-229.
45. Lu, Y. (2019). Artificial intelligence: a survey on evolution, models, applications and future trends. Journal of Management Analytics, 6(1), 1-29.
46. Milovanović, S., Bogdanović, Z., Labus, A., Despotović-Zrakić, M., & Mitrović, S. (2022). Social recruiting: an application of social network analysis for preselection of candidates. Data Technologies and Applications, 56(4), 536-557.
47. Morandini, S., Fraboni, F., De Angelis, M., Puzzo, G., Giusino, D., & Pietrantoni, L. (2023). The impact of artificial intelligence on workers’ skills: Upskilling and reskilling in organisations. Informing Science, 26, 39-68.
48. Natale, S. (2021). Deceitful media: Artificial intelligence and social life after the Turing test. Oxford University Press.
49. Oksanen, R. (2018). New technology-based recruitment methods. University of Tampere Faculty of Management.
50. Ore, O., & Sposato, M. (2022). Opportunities and risks of artificial intelligence in recruitment and selection. International Journal of Organizational Analysis, 30(6), 1771-1782.
51. Otoo, I. C., Assuming, J., & Mensah Agyei, P. (2018). Effectiveness of recruitment and selection practices in public sector higher education institutions: Evidence from Ghana.
52. Pan, Y., Froese, F., Liu, N., Hu, Y., & Ye, M. (2023). The adoption of artificial intelligence in employee recruitment: The influence of contextual factors. In Artificial intelligence and international HRM (pp. 60-82). Routledge.
53. Papajorgji, P., & Moskowitz, H. (2024). Introduction to Artificial Intelligence. In The Mind of Everyday: Combining Individual and Artificial Intelligence (pp. 67-89). Cham: Springer Nature Switzerland.
54. Peterson, C. L. (2024). Globalisation and Pandemic Management: Issues and Outcomes from COVID-19. Taylor & Francis.
55. Priyadarshini, I., & Cotton, C. (2021). Intelligence in cyberspace: the road to cyber singularity. Journal of Experimental & Theoretical Artificial Intelligence, 33(4), 683-717.
56. Raj, R., & Kos, A. (2023). Artificial Intelligence: Evolution, Developments, Applications, and Future Scope. Przeglad Elektrotechniczny, 99(2).
57. Roppelt, J. S., Greimel, N. S., Kanbach, D. K., Stubner, S., & Maran, T. K. (2024). Artificial intelligence in talent acquisition: a multiple case study on multi-national corporations. Management decision, 62(10), 2986-3007.
58. Seth, A. K. (2024). Conscious artificial intelligence and biological naturalism. Behavioral and Brain Sciences, 1-42.
59. Setúbal, M., Conte, T., Kalinowski, M., & Allex Araújo, A. (2024). Investigating the online recruitment and selection journey of novice software engineers: Anti-patterns and recommendations. Empirical Software Engineering, 29(5), 125.
60. Shaddiq, S., & Irpan, M. (2023). Governance of human resources management in the digital era. Journal of Business and Management Studies, 5(3), 80.
61. Shahruddin, S., & Husain, S. H. (2024). Navigating paradoxes of identity and leadership in the age of digital transformation of construction industry: Architects’ experiences and perceptions. Construction Management and Economics, 42(7), 591-609.
62. Sjøvaag, H. (2024). The business of news in the AI economy. AI Magazine, 45(2), 246-255.
63. Sollohub, D. (2019). Millennials in architecture: Generations, disruption, and the legacy of a profession. University of Texas Press.
64. Suhonen, E. (2025). Integration of AI into HR: An analysis on employee attitudes towards AI chatbots in HR advisory services.
65. Symons, J., & Abumusab, S. (2024). Social agency for artifacts: Chatbots and the ethics of artificial intelligence. Digital Society, 3(1), 2.
66. Tiwaari, Y. J., Yadav, O. S., Yadav Sonali, S. R., & Singh, A. A. Y. (2025). Impact of Artificial Intelligence in Space of Digital Marketing. International Journal of Advanced Engineering, Management and Science, 11(2), 610914.
67. Tracy, S. J. (2024). Qualitative research methods: Collecting evidence, crafting analysis, communicating impact. John Wiley & Sons.
68. Tulgan, B. (2015). Bridging the soft skills gap: How to teach the missing basics to todays young talent. John Wiley & Sons.
69. Vardi, S., & Collings, D. G. (2023). What’s in a name? Talent: A review and research agenda. Human Resource Management Journal, 33(3), 660-682.
70. Vassallo, M., Sattin, D., Parati, E., & Picozzi, M. (2024). Problems of Connectionism. Philosophies, 9(2), 41.
71. Vecchi, A., Della Piana, B., Feola, R., & Crudele, C. (2021). Talent management processes and outcomes in a virtual organization. Business Process Management Journal, 27(7), 1937-1965.
72. Wakelin, K. J., McAra-Couper, J., & Fleming, T. (2024). Using an online platform for conducting face-to-face interviews. International journal of qualitative methods, 23, 16094069241234183.
73. Wheeler, K., & Van de Haterd, B. (2024). Talent Acquisition Excellence: Using Digital Capabilities and Analytics to Improve Recruitment. Kogan Page Publishers.
74. Zhu, S., Yu, T., Xu, T., Chen, H., Dustdar, S., Gigan, S., … & Pan, Y. (2023). Intelligent computing: the latest advances, challenges, and future. Intelligent Computing, 2, 0006.