Companies often begin AI adoption as a technology project: a platform is selected, training is delivered, and leadership expects employees to work faster while the organization achieves more with fewer resources. In practice, however, AI adoption is not only a software rollout. It changes how work is distributed, how quality is controlled, how employees learn, where responsibility sits and what managers are expected to do.
This analysis is based on a research article by Julia Shin and Sandra J. Sucher, published in Harvard Business Review, which draws on 18 semi-structured interviews with partners, managers and junior consultants at two major consulting firms. Consulting and research-intensive organizations are not a direct proxy for every business. They are an important exception because their work depends heavily on analysis, client reports, presentations, knowledge processing and managerial judgment. That is why the impact of AI becomes visible there faster and more sharply. But this exception reveals a trend that is likely to appear gradually in other knowledge-based sectors as well – finance, marketing, education, law, HR, IT, media and public services.
BTU researchers assess that the main lesson for Georgian companies is clear: when adopting AI, it is not enough to ask which tool to buy. More important questions are: who checks AI-generated work, who teaches teams how to use AI responsibly, who defines quality standards, who evaluates risk and who carries responsibility when something goes wrong? In many organizations, these questions fall onto middle managers.
The research shows that AI adoption succeeds or fails not only because of the quality of the technology, but because of the strength of the organization’s middle layer. Senior leaders often see AI as a strategic opportunity: faster service, lower costs, new products and leaner teams. Junior employees often experience immediate productivity gains: text, research, presentations or analysis can be produced much faster. But between these two levels stands the manager, who must review AI output, detect errors, coach employees and maintain quality.
For Georgian business, this is a practical issue. If a company gives employees access to AI tools but does not define rules of use, quality criteria, review processes and accountability, speed may become misleading. Work may be created faster, but correcting it, verifying it and adapting it to the real task may fall heavily on managers.
This also matters for citizens, because AI will increasingly affect the quality of services they receive – in banking, e-commerce, universities, insurance companies, clinics, public services and professional advisory work. If AI output is not checked, the customer may receive a fast but inaccurate answer. For business, this means that the real value of AI is not automation alone. The real value is reliable, verified and usable output.
The central question is therefore not simply whether AI reduces work. It is whether AI shifts the burden to another part of the organization. If companies do not strengthen middle management, AI may become not a source of productivity, but a new form of stress: more checking, more coaching, more uncertainty and more responsibility for people who are already in one of the most difficult organizational positions.
In everyday Georgian business practice, this can appear in simple ways. An employee uses AI to prepare a report quickly, but the manager must verify the facts. A marketing team uses AI to create campaign ideas, but the team lead must decide whether they fit the brand’s voice. A finance team receives a fast summary, but someone must check the logic. HR uses AI to draft text or support assessments, but managers must protect the human and legal context. This is where AI adoption becomes not only a question of tools, but of management culture.
Main analysis
The first major mistake in AI adoption is seeing it only as technology. Organizations often assume that buying a license, opening access to employees and providing a short training session will be enough. But AI begins to have real impact when it enters daily work: emails, analysis, client materials, internal reports, presentations, research, service delivery, code, marketing and decision preparation.
At that point, technological change becomes organizational change. New questions emerge: what can be delegated to AI? What must remain with humans? When is AI output good enough? Who checks sources? Who owns the error? How should a junior employee learn professional judgment if they can rely on ready-made AI output from the first day?
Many of these questions land on the desk of the middle manager. The manager is expected to be a team leader, quality controller, trainer, AI-use interpreter, risk detector and executor of leadership expectations at the same time. If the organization does not strengthen this role in advance, the manager starts doing the work that the system itself should support.
The research uses an important term: “workslop.” It refers to AI-generated work that looks professional on the surface but is weak in substance, does not solve the actual task or requires additional managerial review. This is one of the most important organizational risks of the AI era. The text may look polished, the presentation may look structured, the analysis may sound convincing, but underneath there may be superficiality, a wrong assumption, a missing fact or a misunderstanding of the client’s actual problem.
This risk is especially high in sectors where the quality of work is not measured only by format. Consulting and research-intensive organizations are a particularly useful example because their output is often judgment, analysis, structure and recommendation. But the same logic applies to banks, insurance companies, universities, marketing agencies, law firms, technology companies and public organizations. Wherever AI produces text, analysis or a basis for decision-making, managerial review becomes essential.
A second problem is informal learning. Many teams learn AI use on their own: who writes better prompts, who checks answers better, who finds a useful workflow. If this knowledge is not stored and shared across the organization, each team repeatedly solves the same problems. Time is saved in one task but lost again through duplicated experimentation.
This is why AI adoption requires more than training. It needs an internal knowledge system: where good examples are stored, who collects successful use cases, how teams document their experience, how new employees know where to find rules and how managers see what other teams have already tested.
A third problem is misaligned incentives. If an organization continues to reward only fast delivery, individual output or utilization rates, employees will have little motivation to share good AI practices with others. Managers will also lack the time and incentive to coach their teams, because daily delivery always appears more urgent.
Successful AI adoption requires organizations to value not only the final output, but also knowledge sharing, quality control, employee development and better work processes. If a manager teaches the team how to use AI well, checks risks, creates reusable templates, collects effective prompts, defines review rules and makes this knowledge available to the organization, that should count as real managerial value.
A fourth problem is the gap between leadership and managerial reality. Senior leaders often see AI as a strategic opportunity, but they may see less of the everyday friction: how long it takes to check AI output, how often systems make mistakes, how junior employees lose parts of traditional professional learning, how teams feel uncertain about acknowledging AI use, or how clients interpret work that may have been AI-assisted.
This gap becomes especially risky when leadership expects AI to produce fast savings. If managers do not have time, standards and support, they begin to decide each case individually. One team sets one rule for AI use, another team sets a different rule, a third team hides AI use, while a fourth trusts the system too much. The result is inconsistent quality.
For Georgian companies, the lesson is straightforward: AI-use rules should not be left only to the interpretation of individual managers. The organization must define what types of work can involve AI, where human review is required, how AI involvement should be documented, what information should not be entered into models, how risk should be assessed and what employees should do when AI output is questionable.
Visible leadership engagement is especially important. Leaders should not participate only in presentations about AI. They should also join working sessions, listen to managers, observe where AI workflows get stuck and understand the real cost of quality control. When managers are left alone, they carry the true cost of AI transformation while leadership sees only high-level progress.
Another important issue is the leadership pipeline. Traditionally, junior employees learned by observing managers closely: how to structure a workplan, test an analysis, write a recommendation, handle a difficult client conversation and turn weak material into a strong conclusion. Hybrid work has already weakened parts of this apprenticeship. AI can weaken it further if junior employees receive ready-made output without learning how professional judgment is built.
This creates a serious risk: AI may accelerate junior output while weakening the development of future leaders. A junior employee may produce a polished presentation quickly but fail to learn how to evaluate the underlying analysis. They may receive a fast conclusion but not learn how to form their own professional judgment.
That is why the goal of AI adoption should not be speed alone. It should be better organizational learning. AI should support junior employees, but it should not remove the path of professional development. Managers should be freed from repetitive checking where possible so that they can spend more time coaching, explaining quality standards and developing future leaders.
In Georgia, this is especially important for companies that are growing quickly but still have weak management systems. AI can become an opportunity if it is introduced together with stronger processes, internal knowledge bases, standards and management training. But if AI is simply added to chaotic workflows, it will multiply the chaos faster.
Georgia in context: a few data points
Julia Shin and Sandra J. Sucher’s research is based on 18 semi-structured interviews with partners, managers and junior consultants at two major consulting firms. It shows that the greatest operational pressure of AI adoption often falls on middle managers.
International research also shows that many organizations already use AI in at least one business function, but only a smaller share have developed the capabilities to create measurable value beyond initial pilots. This gap suggests that the problem is often not tool access, but weak workflow design, organizational support and management systems.
The research highlights workflow redesign as the key source of AI impact. In other words, AI creates value when work itself is redesigned, not when a new tool is simply added to old processes.
The burden on middle managers also connects to a broader problem: in many organizations, middle managers were already overloaded due to restructuring, leaner teams and wider spans of supervision. AI did not create the problem, but it accelerates it.
For Georgia, additional local research is needed on how companies use AI, who reviews AI output, what role middle managers play, whether internal AI rules exist and how managerial workload changes after AI adoption.
Why this matters for Georgia
For Georgia, AI adoption can become an important source of productivity growth, especially for small and medium-sized businesses, finance, education, marketing, IT services, media and public services. But that result will not happen automatically.
If companies use AI only as a tool for faster text, presentations or analysis, without defining quality rules, the result may be superficial efficiency. Work will look faster on the surface, but the hidden burden of review, error correction and accountability will grow.
This matters especially because middle management is already stretched in many Georgian organizations. A manager is often an operational leader, client communicator, team coach, problem solver and person responsible for results at the same time. If AI oversight is added to all of this without support, the role becomes even heavier.
For the public sector and large organizations, the lesson is broader: AI adoption must be planned as a human and organizational transformation. Technology creates value only when accompanied by clear accountability, data rules, quality control, learning systems and management support.
For education, this means that management programmes should teach AI not only as a technological tool, but as an organizational change process. Future managers need to understand how to review AI output, redesign AI-supported workflows, protect teams from overload and preserve human learning under automation.
What Georgian businesses should do
The first step is defining AI-use rules. A company should know which tasks can use AI, where human review is mandatory, what data cannot be entered into models, how AI involvement should be disclosed and who carries final responsibility.
The second step is targeted support for middle managers. They need more than general AI training. They need practical skills: how to detect hallucinations, how to evaluate prompt quality, how to verify AI-generated analysis and how to teach junior employees to use AI without losing professional judgment.
The third step is creating an internal knowledge base. Companies should collect successful use cases, workflows, useful prompts, review rules, examples of mistakes and team-level learning. This knowledge should not remain scattered across individual computers or chat histories.
The fourth step is changing incentives. If an organization rewards only speed, employees and managers will push knowledge sharing aside. Successful AI adoption should reward coaching, team development, knowledge transfer, quality protection and the creation of internal standards.
The fifth step is real leadership engagement. Senior leaders should listen to middle managers’ experience, see where AI workflows get stuck, understand how much time review takes, notice what junior employees are missing and clarify the standards teams need.
The sixth step is protecting the leadership pipeline. AI should not become a reason why junior employees stop learning professional thinking. Companies need to preserve apprenticeship: learning from senior colleagues, testing analysis, critical thinking, client communication and the ability to distinguish strong work from weak work.
BTUAI assessment
BTUAI assesses that the main risk of AI adoption for Georgian business is not only choosing the wrong technology. A larger risk is failing to see where the real workload goes – and allowing it to fall invisibly onto middle managers.
If a company adopts AI without redesigning workflows, creating quality standards, supporting managers and protecting learning, AI may become a new form of bureaucracy: fast, polished and often in need of review, correction and interpretation.
BTUAI assesses that successful AI transformation begins not with the tool, but with management architecture. A company should know how work moves between humans and AI, how employee roles change, how managerial responsibility grows, how knowledge is stored and how real value is measured.
For Georgia, the opportunity is significant. AI can help companies improve service, reduce costs, accelerate analysis and make better decisions. But this requires strengthening middle management. This is the layer that turns AI ambition into everyday results.
Key findings
AI adoption is not only a technology process. It changes work distribution, accountability, quality control and learning systems.
Julia Shin and Sandra J. Sucher’s research is based on 18 semi-structured interviews with partners, managers and junior consultants at two major consulting firms.
Consulting and research-intensive organizations are not a direct analogy for every business, but they clearly reveal how AI pressure appears in knowledge-based work.
The greatest operational pressure of AI adoption often falls on middle managers.
AI-generated work that looks polished but is weak in substance creates additional managerial burden.
The real value of AI depends on workflow redesign, internal knowledge sharing and quality standards.
Georgian companies need AI-use rules, targeted manager training and internal knowledge systems.
If companies do not strengthen middle management, AI may become organizational overload instead of productivity growth.
Data and evidence base
The Harvard Business Review article by Julia Shin and Sandra J. Sucher is based on 18 semi-structured interviews with partners, managers and junior consultants at two major consulting firms.
The purpose of the research was not to measure broad attitudes, but to understand how people at different organizational levels actually use AI, what support they receive and where friction emerges.
The article notes that many organizations already use AI in at least one business function, but only a smaller share generate measurable value beyond pilots. This shows that AI use and value creation are not the same thing.
The research highlights workflow redesign as the key driver of AI impact. If the process does not change, AI simply adds a new tool to an old organizational logic.
The article describes three major breakdowns: learning is often informal while delivery pressure remains constant; incentives do not reward the behaviours needed for successful AI adoption; and leaders and managers often operate in different realities.
For Georgia, additional local research is needed: how many companies use AI in daily work, how many have AI-use policies, what role middle managers play, how their workload changes and how AI’s real productivity effect is measured.
Methodology
This report was prepared as part of BTUAI Research. The analysis is based on demographic, regional, economic and behavioral data, as well as general trends observed in publicly available sources. The materials are processed using analytical methods applied by BTU researchers, with the support of BTUAI.
The purpose of the research is not to provide personal assessments, but to identify broader trends and practical directions for business, education and society.
This article uses the research analysis by Julia Shin and Sandra J. Sucher, published in Harvard Business Review, on how AI adoption increases pressure on middle managers. The international research is adapted to the context of Georgian business, knowledge-based work, management and AI transformation.
Limitations
This material is analytical and educational in nature. It does not constitute management, legal, financial or HR consulting advice for a specific organization. Specific decisions should be made with the involvement of relevant professionals.
The international research is based on two major consulting firms. This environment is not a direct analogy for every business. The article uses this example as a clear indicator of how AI affects knowledge-based work.
Detailed public data on AI adoption, middle-management workload and AI productivity effects in Georgia remains limited. Additional local research is needed.
Sources
Harvard Business Review – Julia Shin and Sandra J. Sucher, “AI Adoption Is Overloading Your Middle Managers,” 2026.
Julia Shin and Sandra J. Sucher’s research analysis based on 18 semi-structured interviews with partners, managers and junior consultants at two major consulting firms.
International research on AI adoption, workflow redesign, middle management, leadership pipeline and organizational transformation.
BTUAI Research Team – Georgia-focused analytical interpretation.
FAQ
Why does AI overload middle managers?
Because reviewing AI-generated work, detecting errors, coaching employees, protecting quality and meeting leadership expectations often fall onto middle managers.
Isn’t AI supposed to reduce work?
AI can reduce certain tasks, but without standards and review processes it creates new work – especially for those responsible for quality.
What is “workslop”?
It is AI-generated work that looks professional on the surface but is weak in substance, superficial or not useful for the actual task.
What should a Georgian company do?
It should define AI-use rules, create review standards, support managers, build internal knowledge systems and measure AI value by real productivity, not by usage alone.
Why is strengthening middle management important?
Middle managers turn leadership strategy into daily work. If this layer is overloaded, AI benefits cannot scale across the organization.
What is the risk for junior employees?
If AI produces polished work too quickly, junior employees may learn less about professional judgment, analysis review and quality assessment. This weakens the future leadership pipeline.
Keywords
AI adoption; middle managers; Georgian business; AI and management; artificial intelligence in organizations; AI transformation; workflow redesign; managerial overload; AI productivity; knowledge work; leadership development; BTUAI; Business and Technology University; organizational transformation; responsible AI management.
Citation format
BTUAI Research Team. “AI Adoption Is Overloading Middle Managers: What Georgian Business Should Consider.” Business and Technology University, BTUAI.ge, 2026.
Authorship and BTUAI standard footer
Prepared by the academic team of Business and Technology University and the BTUAI Research Team.
Tbilisi, Georgia
BTUAI is an analytical platform of Business and Technology University that studies the impact of artificial intelligence, digital transformation, innovation, startup ecosystems, data analytics and emerging technologies on business, the economy, education and society. BTUAI materials are designed to explain complex technological and economic changes in a clear, reliable and Georgia-focused way.