Artificial intelligence is no longer just another tool sitting in the technology department. It is finding its way into recruitment, daily workflows, decision-making and employee management, creating opportunities for businesses and fresh questions for their leaders.
The upside is easy to see. AI can process large volumes of data, automate repetitive tasks, support faster decisions and free employees to focus on more strategic work. A survey cited in AIHR’s analysis of AI in HR found that 65% of HR professionals reported a positive impact on departmental productivity and efficiency.
But introducing AI is not as simple as switching on a new tool and waiting for productivity to soar. Organisations must also work through skills gaps, integration costs, data security, employee concerns and questions about accountability.
Research from Esade, AIHR and the University of Oxford’s Institute for Ethics in AI offers different perspectives on these changes. Together, the material points to eight emerging workplace practices that are reshaping management, from employees experimenting with AI tools to organisations rethinking how people and automated systems work together.
1. Employees are bringing their own AI tools to work
Employees do not always wait for an official technology rollout before experimenting with AI. They may turn to publicly available tools to draft emails, summarise information, organise research or speed up routine tasks.
This can help employees discover practical uses for AI and reduce the time spent on repetitive work. However, when employees use tools that their organisations have not approved, questions arise about data protection, security and oversight.
The supplied material on AI adoption identifies this behaviour as the bring-your-own-AI (BYOAI) movement. It also refers to claims that employees may conceal their AI use because they worry about how managers will perceive it. The specific percentages supplied for BYOAI adoption and undisclosed usage could not be verified from the accompanying sources, so they should not be treated as established findings.
The management challenge: Organisations need to distinguish useful experimentation from risky or inappropriate tool use.
Managers can help by:
Establishing clear guidance on which AI tools employees may use.
Explaining what kinds of company, customer and employee data must not be entered into unapproved systems.
Providing secure alternatives for common workplace tasks.
Encouraging employees to discuss their AI use without automatically treating experimentation as misconduct.
A blanket ban may limit exposure to certain risks, but it can also make legitimate use harder to discuss. Clear rules and approved tools give employees a more practical way forward.
2. Creating content is giving way to reviewing it
Generative AI can produce first drafts, summarise documents, organise information and assist with routine formatting. Employees can use these capabilities to move through some tasks faster, although the results still require checking.
As AI takes on parts of content creation, the human contribution may shift towards assessing accuracy, refining language, checking context and deciding whether an output is appropriate for its intended purpose.
Esade’s analysis of AI in business identifies automation and improved decision-making among the technology’s potential benefits. AIHR similarly highlights the importance of interpreting AI-generated insights rather than simply knowing how to operate a tool.
The management challenge: Performance assessment needs to account for the quality of human judgement, not just the quantity of work produced.
For managers, this could mean examining:
Whether employees can identify inaccurate or misleading AI-generated information.
How effectively they adapt outputs to business needs.
Whether their final work demonstrates sound judgement and subject knowledge.
How AI-assisted work contributes to measurable team or business outcomes.
The change does not make productivity measurement irrelevant. It makes the choice of productivity measures more important, particularly when the first version of a document or analysis can be generated quickly.
3. AI agents are expanding the question of accountability
AI assistants generally respond to requests, while more autonomous AI agents can be designed to carry out sequences of tasks towards a defined goal. Depending on their permissions and configuration, they may interact with software, retrieve information and move work between different stages of a process.
Such systems could support activities that involve multiple routine steps. However, the supplied material does not establish how widely autonomous agents have been deployed across organisations or how much managerial work they currently perform.
The distinction matters because a system that executes several actions can create different oversight requirements from a tool that simply drafts a response.
The management challenge: Organisations must decide who is responsible when an AI system makes an error or takes an inappropriate action.
A practical approach involves:
Defining the tasks an agent may perform and the boundaries it must respect.
Restricting access to sensitive systems and information.
Establishing when human approval is required.
Maintaining records that allow decisions and actions to be reviewed.
Assigning a human owner who is responsible for monitoring the system.
AI agents do not need to be treated as human employees to require clear accountability. Their permissions, responsibilities and limits need to be defined before organisations rely on them for consequential work.
4. Middle managers may need to rethink their role
AI can automate or assist with administrative activities such as summarising discussions, organising information and preparing reports. Esade identifies process automation as a potential source of efficiency, while AIHR highlights the need to integrate tools into existing workflows rather than introducing isolated solutions.
These capabilities could reduce the time managers spend gathering and relaying information. However, the available sources do not establish that AI is making middle-management roles redundant across organisations.
The more useful question is how managerial responsibilities might change when some administrative tasks become easier to automate.
The management challenge: Organisations must decide how to use the time saved without losing coordination, coaching and accountability.
Managers may need to devote greater attention to:
Helping teams interpret information and make decisions.
Coaching employees and supporting professional development.
Resolving conflicts and coordinating work across functions.
Identifying where AI improves a process and where human intervention remains necessary.
If a manager's day becomes less crowded with routine reporting, the opportunity is to spend more time on work requiring judgement and relationships. Whether this happens depends on how organisations redesign roles, rather than on automation alone.
5. AI fluency is becoming a workforce development priority
Using AI effectively involves more than knowing which buttons to press. Employees need to understand the capabilities and limitations of the tools they use, assess the quality of their outputs and recognise when a task requires human expertise.
AIHR identifies skill gaps and resistance to change as important barriers to AI adoption in HR. Its analysis also stresses the need for training that helps employees interpret AI-generated insights and understand where algorithms can fail.
Esade similarly recommends training personnel as part of effective AI implementation.
The management challenge: Employers must connect AI training to the work employees actually do.
Rather than relying exclusively on general introductions to AI, organisations can:
Identify the tasks in which each team could use AI.
Provide role-specific exercises and practical guidance.
Teach employees how to verify outputs and protect sensitive information.
Give teams opportunities to test tools against real workplace needs.
Assess whether training improves work quality and confidence.
A recruitment team, for example, may need to learn how to evaluate AI-assisted screening without overlooking qualified candidates. A finance team may focus on checking summaries and calculations. The tools may overlap, but the skills required to use them responsibly will differ.
6. Faster workflows do not automatically mean lighter workloads
AI can reduce the time needed for certain activities, but the overall effect on employees depends on how work is reorganised. Faster document preparation or data analysis may free time for other tasks, or it may lead managers to increase output expectations.
The supplied brief describes a potential tension between improved job satisfaction for some routine tasks and greater cognitive strain when employees must handle more complex decisions. However, the accompanying sources do not verify the specific claim of a 41% increase in decision-making time, so the figure should not be presented as established evidence.
AIHR's analysis does identify employee wellbeing, human connection and resistance to change as relevant considerations when organisations automate HR activities.
The management challenge: Leaders must assess whether AI is improving the working day overall, rather than judging success solely by the speed of individual tasks.
Managers can examine:
Whether employees are taking on additional work after automation.
How much checking and correction AI-generated outputs require.
Whether employees have sufficient time for complex tasks.
Whether AI use affects job satisfaction, collaboration or access to human support.
Automation can remove repetitive work, but it does not automatically remove pressure. Organisations need to examine the work that replaces those tasks and adjust expectations accordingly.
7. Hiring may place greater emphasis on adaptability
AI can assist with specialised activities, including drafting code, reviewing documents and analysing information. Its capabilities may change the mix of skills required for particular roles, but the extent of that change varies by task, profession and organisation.
AIHR highlights the need for HR teams to develop data literacy, technical understanding and the ability to identify appropriate AI use cases. Esade also points to the emergence of new professional profiles in areas such as algorithms, technology, programming, strategy and maintenance.
The evidence therefore supports a broadening of skills requirements, rather than a universal shift away from specialists towards generalists.
The management challenge: Employers must decide which technical skills remain essential and which complementary capabilities help employees work effectively with AI.
Talent teams may need to consider:
The ability to evaluate information critically.
Adaptability when tools and workflows change.
Communication and collaboration across functions.
The technical knowledge required for each role.
The judgement needed to recognise when AI should not be used.
Specialist expertise remains important in roles where accuracy, technical depth or professional accountability are essential. Adaptability can complement that expertise, helping employees apply their knowledge as tools and processes evolve.
8. Organisations must decide how to use the capacity AI creates
When AI shortens a task or automates part of a process, organisations may gain additional capacity. The business outcome depends on what happens next: teams might use the time for customer service, innovation, analysis or other priorities, while some organisations may reassess staffing and operating costs.
Esade's analysis identifies efficiency, cost reduction and improved decision-making among AI's potential business benefits. It also notes that automation can contribute to job displacement, while creating demand for new professional profiles.
Neither outcome is automatic. The effect depends on the tasks involved, the organisation's strategy and the way work is redesigned.
The management challenge: Leaders need to make deliberate decisions about how to redeploy capacity and communicate those decisions clearly.
Organisations can consider whether the time saved could support:
More responsive customer service.
Projects that teams previously lacked time to pursue.
Employee learning and the development of new capabilities.
Improvements to processes that remain slow or inefficient.
More manageable workloads and better service quality.
Where changes affect roles or staffing, employers also need to communicate honestly with employees. Clear explanations of how AI will affect work can help teams understand the business rationale and prepare for new responsibilities.
What the eight practices mean for management
AI adoption is not simply a choice between embracing automation and avoiding its risks. Organisations need to understand where the technology adds value, what it costs to implement and how it changes responsibilities across the workforce.
The three sources point to several priorities for leaders:
Strategy: Esade recommends identifying suitable use cases, implementing AI gradually and monitoring performance.
Skills: AIHR emphasises training, AI readiness and the ability to evaluate outputs critically.
Governance: The Oxford researchers highlight worker participation, accountability and the implications of algorithmic management for workplace power.
Integration: AIHR notes that new tools must work with existing HR systems and workflows to deliver their intended benefits.
Business value: Esade highlights efficiency, decision support and customer experience as potential gains, alongside costs, privacy concerns and possible job displacement.
These perspectives are complementary, but they are not interchangeable. Business benefits, HR implementation challenges and questions about worker rights represent different parts of the management problem.
The next management question is what happens after adoption
AI can help organisations automate routine tasks, analyse information and improve selected processes. It can also introduce costs, skill requirements and governance questions that determine whether those benefits materialise.
For managers, the task is to connect AI adoption to clear business needs while paying attention to the people expected to use the technology. Success will depend on practical training, appropriate oversight, well-integrated systems and realistic decisions about how work changes.
The most useful measure of progress is not how many AI tools an organisation has adopted. It is whether those tools help employees and busines
