Conceptualization of power in organizations
Beyond the internal dynamics of organisations, the preservation of the value of entry-level jobs is conditioned by a plurality of forms of power that interact with each other. Institutional power, exercised by governments and trade unions, is expressed through employment policies, minimum wages and youth insertion programmes. Economic power, on the other hand, comes from companies, which decide to maintain or automate these jobs based on the criteria of efficiency and profitability. Added to this is the symbolic power, visible in the social perception of these jobs, often considered transitory or low-skilled, which limits their recognition and valorisation. Finally, technological power redefines the tasks considered necessary and the profiles enabled to perform them, directly affecting the stability and future of these roles (ECLAC, 2025; Axial ERP, 2025).
In this context, the emergence of AI introduces profound transformations that directly reconfigure entry-level jobs. By automating routine, administrative, and operational tasks, AI alters the very structure of these positions and shifts the balance of power. It’s advanced impacts institutional power by demanding new training and protection policies aimed at young workers. It also redefines symbolic power, since the automation of functions traditionally considered formative weakens their value as learning stages. According to Infobae (2025), AI could lead to the disappearance of about 50% of entry-level office jobs in the next five years, forcing a rethink of career paths. However, if implemented from an inclusive perspective, AI can also expand access to new skills and emerging roles, thereby expanding development opportunities (ILO, 2025; Forbes Argentina, 2025).
Within companies, these transformations are intertwined with forms of power expressed in both everyday interactions and formal structures. Among colleagues, power is exercised through alliances, prestige, or control over information, generating informal dynamics that can favour cooperation or, conversely, produce exclusion (Edelman et al., 2024). In the hierarchy, legitimate authority grants decision-making power but also generates relationships of dependence and resistance that can slow innovation (Kang, 2022).
In turn, the human resources department exercises normative power through policies and procedures, acting as a mediator although frequently privileging institutional interests over individual ones (Mariño-Arévalo, 2014). These cross-interactions show that power does not reside solely in the formal structure, but also in informal networks and perceived legitimacy, generating diffuse zones of responsibility and professional subjectivities that influence organisational stability (Álvarez Sánchez, 2011; Porras Velásquez, 2018).
From this perspective, it is pertinent to mobilise the major theoretical currents to deepen the understanding of these dynamics. In sociology, Michel Foucault argues that power is diffuse and manifests itself in social relations rather than only in hierarchical structures. In contrast, Max Weber classifies power according to its legitimacy, distinguishing three types: traditional, charismatic, and legal-rational. For his part, Pierre Bourdieu introduces the notions of symbolic, cultural, and social capital as sources of power, thus broadening the understanding of social dynamics beyond the economic and institutional dimensions (Álvarez Sánchez, 2010).
Types of power in organizations
Understanding these transformations requires not only analysing how AI reconfigures labour relations, but also identifying the specific forms that power takes within organisations. In this sense, the classic literature on organisational behaviour distinguishes several types of power that explain how influence is exerted in work environments and how resources, decisions, and opportunities are distributed.
From this perspective, in the organisational sphere, power does not manifest in a single form. Still, it takes different forms depending on its origin, legitimacy, and the mechanisms through which it influences individual behaviour. Among the main types of power identified in the literature are:
- Legitimate power is derived from the formal position a person occupies within a hierarchical structure and is supported by institutional authority (such as that of a director),
- Coercive power is based on the ability to impose sanctions or negative consequences, generating obedience through pressure or fear,
- The power of reward, through the possibility of offering incentives, benefits, or recognition, motivates the achievement of objectives,
- The power of experience and expertise, which comes from technical knowledge, specialised skills and a professional career, grants influence regardless of the hierarchical level,
- The referent power, associated with the charisma, admiration, and respect a person inspires, allowing them to influence through their personal qualities and their ability to build bonds of trust.
Power dynamics in organizational environments
Power is not limited to formal hierarchical structures but manifests itself in a wide variety of dynamics that cut across different levels of interaction. These dynamics take the form of tensions, paradoxes, and conflicts and are expressed through phenomena related to status, normative pressure, deviance, and generational and gender differences (Simeonova, Galliers, & Karanasios, 2018).
From this perspective, the authors point out that organisational power is deployed across multiple areas beyond hierarchy. It becomes visible in the processes of organisational change, in the formulation of strategies, in knowledge management, in coordination and collective learning, in intra- and inter-organisational work, in the formal structure and internal rules, in social networks, in leadership, in interpersonal relationships, in the use and implementation of technologies and in the autonomy of roles.
Artificial intelligence: a new resource or threat to the establishment?
The introduction of AI in work environments invites us to question how it transforms traditionally established forms of power. To understand this reconfiguration, it is useful to mobilise different theoretical frameworks to analyse the legitimacy, domination, and distribution of resources in organisations.
Let’s ask ourselves about the impact of AI on different forms of power. According to Max Weber, there are three types of legitimacy that structure power: traditional, charismatic and legal-rational. In the company, traditional power, based on habit and continuity, is weakened by AI, which introduces a dynamic of rationalisation and innovation that breaks with legacy practices.
Charismatic power, on the other hand, is revealed to be deeply ambivalent: AI can amplify a leader’s aura by enabling them to build a hyperpersonality, but it can also erode it by shifting attention to the algorithms’ strength. A form of technological charisma could even emerge, no longer embodied by the individual, but by the machine itself.
However, legal-rational power finds in AI a privileged terrain, since by automating processes and speeding up decision-making, they take rationalisation to its extreme, with the risk of establishing an algorithmic bureaucracy even more impersonal than the one described by Weber. In this scenario, legitimacy could progressively shift from human institutions to technical systems.
In short, AI disrupts the three forms of power legitimacy identified by Weber: it weakens tradition, redefines charisma, and reinforces bureaucratic rationalisation. It opens the way to a new configuration in which the machine could become a source of legitimacy. For companies, the challenge will be to preserve the balance between algorithmic efficiency and human responsibility.
If we broaden our gaze to Pierre Bourdieu’s thought, power is understood as a set of relations of domination inscribed in social structures, where individuals have different capitals, such as economic, cultural, social and symbolic, that determine their position and influence. From this perspective, AI tends to strengthen economic capital by generating new sources of profit (data, algorithms, automation), consolidating the dominance of already powerful actors, and marginalising those without access to these resources.
Its impact on cultural capital is also significant, although it democratises certain knowledge, widening the gaps between those who master these tools and those who are excluded. Digital skills thus become a new form of cultural capital, valued in the labour market and recognised socially.
Social capital is also transformed, as through social networks, AI facilitates the creation of virtual communities but makes links more dependent on algorithmic logics. Power is then concentrated in those who know how to optimise their visibility through digital tools, while others risk being isolated. Social capital, mediated by algorithms, is transformed by digital visibility, while symbolic capital is displaced towards classification and recommendation systems that confer prestige and legitimacy through opaque technical logics. Together, AI reconfigures the capitals described by Bourdieu, amplifying inequalities and generating new forms of domination linked to technological dominance.
After this sociological approach, it is pertinent to return to the organisational terrain to analyse how AI affects the forms of power classically identified in the literature of organisational behaviour: legitimate, coercive, reward, experience, and referent power:
- Legitimate power: AI can strengthen institutional authority by making decisions more rational and credible. However, this power can be weakened if employees perceive the algorithm as the true authority, to the detriment of the hierarchical superior,
- Coercive power: AI facilitates control by monitoring performances and behaviours, reinforcing the ability to sanction. But this permanent surveillance can generate stress and a sense of dehumanisation in employees,
- Reward power: AI allows incentives to be customised based on performance data, making rewards more objective, but potentially less human and symbolic,
- Power of expertise: AI reduces reliance on certain experts by automating complex tasks while valuing those proficient in digital tools. The power of experience is then shifted to the ability to use AI effectively,
- Referent power: AI can amplify the charisma of the leader but can also compete with it: admiration and respect can be transferred to those who master AI and even to the machine itself, with the risk of reducing human warmth in labour relations.
In short, the rise of AI profoundly disrupts power relations within organisations. It promises greater efficacy and objectivity, but it also poses significant risks of dehumanisation and loss of human legitimacy. The real challenge will be to maintain a balance between technological authority and the relational dimension, which is essential for collective work. More broadly, AI redefines legitimacy and accentuates the dominance of those who shape its codes. In the face of these mutations, it is crucial to interrogate the new emerging powers, in particular those of algorithmic manipulation, where algorithms now guide behaviours and decisions.
Emerging Powers in Algorithmic Manipulation (AM)
The analysis of algorithmic manipulation reveals the emergence of new forms of power that reconfigure social and professional relations. It is no longer the introduction of AI itself that transforms organisational dynamics, but the ability of algorithms to intervene discreetly, continuously, and highly effectively in decision-making processes. As these systems are integrated into the daily life of organisations, four dimensions of power emerge that redraw traditional balances (Zuboff, 2019; Susskind, 2020):
- Technical power, wielded by those who design and adjust algorithms, determines what data is processed and how it is interpreted,
- Structural power occurs when algorithms set invisible rules that affect day-to-day decisions (e.g., task assignment),
- Symbolic power, which confers legitimacy on algorithms by presenting them as objective and neutral, even if they reproduce human biases,
- Political power, which influences collective processes such as elections or public policies (Zuboff, 2019; Susskind, 2020),
- These dimensions do not act in isolation but are intertwined within a power ecosystem profoundly transformed by algorithmic logic.
Technical power: the authority of those who control the code
Technical power is exercised by data scientists, AI engineers, and system architects who decide which data to select, how to interpret it, and which models to use to process it. Their mastery of computational logics gives them a strategic, though often invisible, authority in the formal hierarchy. Operational managers, responsible for HR, and HR and employees rely on mechanisms they don’t fully understand, leading to cognitive and decisional asymmetry. Technical authority thus tends to replace hierarchical authority, shifting power to those who control the algorithm’s design.
Structural Power: Invisible Rules That Shape Action
Structural power is migrating to AI project managers, business leaders, and digital transformation leaders. By integrating algorithms into everyday processes such as task assignment, scheduling, and file prioritisation, these actors establish implicit rules that are rarely debated but deeply normative. For field teams, this automation reduces autonomy and standardises practices. Algorithmic logics impose rhythms, thresholds and standards without the need for explicit human authority, configuring an invisible architecture that guides behaviours.
Symbolic power: the belief in machine neutrality
Symbolic power is expressed through the legitimacy conferred on algorithms. The general management and innovation managers present AI as objective, neutral, and rational, which enables them to justify strategic decisions and neutralise opposition. Trade unions, staff representatives and employees find it difficult to challenge decisions that, although derived from biased models, are perceived as unquestionable. The apparent objectivity of the algorithm masks political and economic arbitrage beneath a veneer of technical rationality. Symbolic power thus rests on a widely shared belief: the machine tells the truth.
Political Power: Data-Driven Governance
In a context of strategic uncertainty, political power shifts to executive committees, strategic management, and shareholders who use algorithms to pilot restructurings, guide investments, or model returns. AI becomes a government instrument capable of simulating scenarios and rationalising decisions. This shift towards data weakens internal counter-powers: middle management, work collectives and social partners see their capacity for influence reduced. Algorithmic manipulation thus imposes itself as a discreet but highly effective form of centralisation, which redefines the modalities of collective action.
These forms of power, although abstract in their formulation, acquire a tangible consistency when observed in concrete devices. Algorithmic recruitment is a paradigmatic example: it simultaneously crystallises technical power (the selection of variables), structural power (filtering criteria), symbolic power (the algorithm’s legitimacy), and political power (strategic decisions about profiles). This case demonstrates the differentiated effects of algorithmic manipulation across the company’s various actors.
Fracture of competences and inequities in decision-making
we evolved in an algorithmic universe that restricts access to alternative visions of reality. In addition, algorithms tend to produce homogeneous, conservative, repetitive, and redundant results (Innerarity, 2025).
In this scenario, the recruitment of young talent highlights the intertwining of technical, symbolic, structural and social dimensions. The rise of AI disrupts these dynamics by automating procedures, redefining selection criteria, and shifting the balance of power between candidates and recruiters. A significant fracture in decision-making processes has emerged: the accelerated reconfiguration of the labour market by AI is causing massive worker displacement across sectors. For young graduates, this evolution is a major challenge, as they face a double competency: that of other applicants and that of increasingly efficient automated systems. Specifically, a medium-sized company that previously relied on an HR team to manually sift through hundreds of resumes now uses an AI system to filter applications. This system analyses CVs, identifies keywords, evaluates compatibility with the profile sought and can even conduct preliminary interviews using chatbots.
From this perspective, AI has double potential. On the one hand, it can reduce gaps by freeing less skilled employees from repetitive tasks, facilitating access to intelligent tools, and offering personalised training. On the other hand, it risks expanding them if their deployment is not equally framed. The document identifies several biases that complicate the hiring, promotion, and recognition processes: academic degree bias, status bias, affinity bias, as well as gender, age, and confirmation biases. These mechanisms reinforce invisible hierarchies and consolidate power asymmetries, making it difficult to build inclusive and equitable work environments (Álvarez Sánchez, 2011; Mariño-Arévalo, 2014; Porras Velásquez, 2018).
The European Parliament’s report (2020) summarises this ambivalence by highlighting both the opportunities and risks of AI. Among the former are the automation of repetitive tasks, access to intuitive interfaces, and personalised training. Among the risks are labour displacement, knowledge concentration, and the absence of inclusive policies. On this basis, the paper raises a key question: how does AI’s impact on forms of power relate to the evolution of skills splitting? The answer calls for analysing how AI redistributes authority, autonomy, and decision-making power within organisations.
Potential for bridging gaps:
- Automating repetitive tasks: AI can free lower-skilled employees from routine tasks, allowing them to focus on more strategic or creative roles,
- Access to intelligent tools: Intuitive interfaces and virtual assistants can help workers with little technical background to use complex systems,
- Personalised training: AI enables identifying skills gaps and proposing tailored training itineraries, thereby helping to democratise learning in the company.
Risks of widening gaps:
- Displacement of the workforce: automation can make certain jobs obsolete, especially those that do not require high qualifications, generating unemployment or precariousness,
- Concentration of knowledge: workers with advanced digital skills are increasingly valued, while others risk marginalisation,
- Absence of inclusive policies: Without training and adaptation strategies, AI can accentuate existing inequalities.
unskilled employees? Thus, it is essential to examine how AI’s impact on power can influence, and even accentuate, the skills fracture between skilled and unskilled employees, especially concerning organisational dynamics and decision-making processes.
The rise of AI is profoundly transforming the professional landscape. What was once considered a simple technological progress is now revealed as an engine of social, organisational and human recomposition. As AI is integrated into business processes, it redefines the boundaries of competencies, creates new opportunities, and accentuates certain imbalances. Intelligent automation doesn’t just replace repetitive activities; it changes the value placed on certain expertise. Employees who master or interpret these technologies see their roles evolve, while those who are not prepared risk losing influence or recognition. As the document points out, “the fracture of competences does not disappear: it is displaced”.
This tension is especially visible in multigenerational teams. A young collaborator familiar with AI can destabilise experienced colleagues, generating both innovation and friction. However, the lack of business knowledge in these emerging profiles reminds us that technological competence alone is not enough. Similarly, integrating a young person without AI skills into a team where the tool becomes central poses new organisational challenges.
In parallel, the modes of supervision and work management are also evolving. AI enables dynamic task assignment, more accurate monitoring, and continuous performance evaluation. At first glance, this may seem to reinforce equity. But in practice, this algorithmic control sometimes accentuates the perception of a gap between those who know how to use these systems intelligently and those who suffer from them. In other words, access to tools becomes a new axis of power.
It is precisely at this point that AI undergoes one of its most significant changes, redistributing capabilities and redefining autonomy. Access to intelligent tools allows each person to overcome their traditional limits, opening the door to understanding, synthesis, and analysis skills previously reserved for experts. This democratisation is a game-changer, given that skills are no longer acquired only through training or experience, but also through the ability to leverage advanced tools.
In continuity with this redistribution, AI democratises and personalises access to training. Intuitive interfaces allow non-technical people to perform advanced tasks, while individualised learning paths lower barriers to entry. However, this democratisation does not eliminate the gap: it reconfigures it. Those who adopt these tools advance faster; those who do not are left behind. This facilitated access to up-to-date skills is directly connected to labour market expectations. Individuals who use AI develop knowledge aligned with current needs, increasing their employability and attractiveness.
Finally, the fracture of competences does not disappear under the effect of AI: it changes its nature. It is no longer structured solely around education or experience, but around adaptability, openness to change, and mastery of constantly evolving technologies. For companies, the responsibility is to accompany this transition, creating conditions in which professional experience, technological competence, and human intelligence complement rather than oppose one another. As the document concludes, “the future balance will not depend only on the power of the tools, but on the way in which organizations know how to evolve the skills of each person so that no one is left out of the way”.
Fracture of competencies and forms of power in the company: a game of reciprocal influences
As discussed in the “4.3 Artificial intelligence: a new resource or threat to the establishment?”, ndly influences the distribution of power within companies. But the relationship also operates in the opposite direction, since existing forms of power can amplify or attenuate this fracture. Thus, it configures a circular dynamic in which competencies, technologies and powers intertwine to redefine organisational balances.
The first link in this circular dynamic appears in legitimate power. When some employees develop AI competencies that their managers lack, formal authority can be weakened. The hierarchy loses part of its operational legitimacy if it no longer understands the technological levers that structure the activity. Conversely, AI-competent employees gain influence, sometimes even bypassing the classic decisional chain. But legitimate power also acts on the fracture. If management strategically invests in AI and ensures equitable access to training, it can reduce the gap between employees. On the other hand, if it maintains decisions that exclude certain groups from access to tools or training, the fracture widens. Legitimate power then becomes an amplifier of technological inequalities.
Coercive power is also profoundly affected by the fracture of competencies. AI introduces new monitoring, evaluation, and control tools, increasing pressure on those who are not proficient in these technologies. These employees are more exposed to sanctions linked to insufficient or misinterpreted performance, reinforcing the fear of technological incompetence. In turn, coercive power influences the fracture. When algorithmic supervision is used as a disciplinary lever, it creates an environment of dependency and fear that inhibits initiative and limits learning. This dynamic exacerbates the long-term fragmentation of competencies, turning coercive power into a barrier to the acquisition of new skills.
The power of reward reproduces this logic of reciprocal influence. In a context where AI mastery becomes strategic, financial, symbolic or promotional rewards tend to be concentrated on employees who possess these competencies. This generates a sense of injustice among the less qualified collaborators and reinforces the fracture. However, reward power can also act as a lever for inclusion. A policy focused on learning, experimentation, and equitable progression can bridge the gap. On the other hand, if rewards are reserved exclusively for AI-related performance, those without access to training are excluded, deepening inequality.
AI especially transforms the power of expertise. The emergence of new technological skills shifts the centre of gravity towards knowledge such as analysis, modelling or automation. AI experts gain influence, while experienced but less technophile collaborators see their informal power diminish. When specialists monopolise technical knowledge, they become guardians of competition, increasing dependence on other employees and reinforcing the fracture. But if expertise is shared through mentoring, internal training or communities of practice, it can become a tool for emancipation. The power of expertise can enclose or release, depending on how it is exercised.
Referent power, based on charisma and the ability to inspire, is also being redefined in the age of AI. Mastering these tools reinforces a leader’s credibility, especially with younger generations for whom technological proficiency is essential. Conversely, leaders who don’t take ownership of AI may see their influence diminish. This power can facilitate the technological transition if it is used to accompany, reassure and motivate. But it can also deepen the fracture if only technophile profiles are valued or the difficulties of those who advance more slowly are minimised.
Together, the skills divide and the forms of power within the firm are closely linked and mutually influence one another. AI acts as a revealer and accelerator of this interdependence: it strengthens certain powers, weakens others, and contributes to the emergence of new implicit hierarchies. In addition, it redefines the value placed on each individual in the organisation.
Understanding these interactions is critical to rethinking internal balances and making AI a driver of inclusion, rather than a factor of inequality. In this perspective, the document recalls that integrating AI-assisted training requires establishing a climate of psychological safety and trust. This environment favours effective learning, stimulates constructive feedback, and facilitates the acceptance of AI-generated recommendations, actively involving employees in the design and implementation of these technologies.
0 Comments