AI and the Recalibration of Work
Public Essay · Artificial Intelligence, Work, and Institutional Trust
Public scholarship · Artificial intelligence and organizational transformation
The central leadership challenge is not adopting a new tool. It is redesigning work as technology, data, public expectations, policy, and human agency change together.
William W. Towns, Ph.D., MBA
September 2026
The debate over artificial intelligence and work is often reduced to a single question: When will AI replace all the workers?
The anxiety is understandable. Job seekers are navigating longer, less personal application processes, while new graduates are entering a market in which employers are reconsidering entry-level work and experimenting with AI. Those experiences deserve attention. They do not, however, tell us by themselves how AI will reshape employment across the economy.
But the dominant view remains too narrow. It treats employment as a contest between a person and a machine. It also assumes that technology changes while everything around it remains still. Neither assumption reflects how institutions or labor markets actually change.
AI is arriving alongside a much broader reset in how people think about and understand work. Employees are reconsidering where, when, and why they work, while employers are redesigning roles and searching for productivity. Customers and communities expect more institutions than just products or services. Policymakers are testing new rules for privacy, accountability, safety, and discrimination. At the same time, the 2026 Edelman Trust Barometer describes a society retreating toward insularity as institutional trust remains strained.4 The public conversation about AI therefore swings between wonder and fear.
AI is part of a societal recalibration in which technology, institutions, and expectations are moving together, but often in different directions.
My Societal Recalibration Cycle offers a way to understand this moment. The cycle contains five conditions: Data & Insights, Disruptive Technology, Public Perception, Political Shifts, and Optimism. None has a fixed position or sequence. When these conditions advance or shift quickly, established beliefs, practices, and institutions can begin to decouple from established norms. The resulting disruption creates pressure for organizations to adapt or risk being viewed as being behind the times.
AI makes this interaction visible. New capabilities change how organizations collect and interpret data. Those uses shape employee and public perceptions of value, fairness, and risk. Public response influences policy, regulation, adoption, and investment, which in turn affect the technology's development and use. Optimism is not the final stage of this process. It is the connecting condition: the belief in progress, confidence to invest, and agency to act that can move institutions from disruption toward intentional adaptation.
Work changes at the level of tasks
Jobs are bundles of tasks, relationships, judgments, routines, and responsibilities. AI may automate one part of a role, support another, and increase the value of a third. That is why sweeping claims about entire occupations being eliminated can obscure more than they reveal.
Rather than defaulting to a doomsday scenario, I encourage leaders to examine AI at the task level. MIT Sloan Executive Education's Executive Leadership in the Age of Artificial Intelligence makes the case directly through a collection of articles that state: technology changes tasks, and jobs are composed of tasks.5 In practice, a translator may use AI for a first pass while applying human judgment to cultural meaning. A manager may automate reporting and spend more time coaching, interpreting, and setting direction. A researcher may use AI to sort information while retaining responsibility for framing the questions and evaluating the answers.
This distinction matters. If leaders treat AI as a labor-reduction tool, they will begin with head count. If they treat it as a work-design tool, they will begin with the work itself:
Which tasks are repetitive, hazardous, or low value?
Which tasks require context, empathy, accountability, or judgment?
Where can AI improve a person's ability to perform?
Where could reliance on AI weaken expertise or create unacceptable risk?
Where will overreliance on AI reduce our competitive advantage?
How should the time released by automation be used?
The last question is often ignored. Time saved is not the same as value created. An organization gains only when it deliberately redirects that capacity toward better service, deeper analysis, stronger relationships, new products, or more meaningful work.
The five conditions shaping AI and work
Disruptive Technology: capability is moving faster than job design
Generative and agentic AI can draft, synthesize, recommend, coordinate, and increasingly execute sequences of work. These capabilities extend automation into cognitive tasks once thought difficult to codify. They also make it possible to redesign workflows across functional boundaries rather than insert a tool into one person's existing routine.
But capability does not determine value on its own. The MIT Sloan collection identifies a persistent gap between proofs of concept and enterprise strategy. Organizations often launch AI pilots without connecting them to business priorities, data readiness, employee capability, or operating discipline. The result is a lot of motion, but no movement.
The better starting point is identifying a real business problem, not a compulsion to deploy AI for AI's sake. Examine the problem with the people directly affected. Break it into steps, and then break those steps into tasks. Decide where AI can assist, where a human must decide, and where the organization should not use AI at all. Then redesign roles, decision rights, controls, and measures around the new workflow.
Data & Insights: AI quality is an organizational condition
AI systems depend on data, but the leadership issue is larger than data volume. Leaders need to know where information comes from, whether it is complete and appropriate, how it changes, who can access it, and how model outputs are tested once a system is in use.
The OECD's framework for classifying AI systems helps make this practical. It asks leaders to examine five dimensions of a system: people and planet, economic context, data and inputs, the AI model, and tasks and outputs. This shifts governance from abstract principles to a concrete examination of a specific system in a specific setting.1
That context matters. An AI tool used to summarize internal meeting notes does not carry the same stakes as one used to screen job candidates, evaluate employee performance, approve credit, or guide medical decisions. The level of scrutiny should follow the consequences of the decision, the people affected, and their ability to challenge an output.
Public Perception: trust becomes part of performance
Employees do not experience AI as a strategy deck. They experience it through changes in workload, monitoring, opportunity, status, and control. Customers experience it through the quality of a decision and whether they can reach a person when the system gets it wrong.
That makes public perception an operating concern, not a communications afterthought. A technically effective system can still fail if people see it as unfair, intrusive, or designed primarily to remove them from the equation.
The 2022 Blueprint for an AI Bill of Rights offers a useful set of protections: safe and effective systems; protection from algorithmic discrimination; data privacy; notice and explanation; and access to human alternatives, consideration, and fallback.2 The document is nonbinding, but its principles capture a durable expectation: people affected by automated decisions should not be treated as raw material for optimization.
Leaders build trust when they explain where AI is being used, invite workers into job redesign, test for unequal effects, and create a meaningful path to question or correct a decision. Trust grows through the design of the system, not through slogans about responsible AI.
Political Shifts: the rules will continue to move
AI adoption is unfolding inside changing policy and regulatory environments. Privacy rules, labor protections, intellectual-property disputes, sector standards, procurement requirements, and expectations for model accountability will continue to evolve.
Organizations should not wait for a final rulebook. They need governance that can adjust as technology and policy change. That means clear ownership, documented use cases, risk classification, review thresholds, monitoring after deployment, and incident response. Compliance is the floor. The larger responsibility is to decide what the organization will do when a use is legal but inconsistent with its mission, values, or the dignity of the people affected. I believe that every board should assign explicit responsibility for AI oversight. Organizations with extensive or high-risk AI applications should have a dedicated AI committee separate from or a sub set of their technology committee.
Optimism: agency turns disruption into institutional action
Optimism sits at the center of the Societal Recalibration Cycle because change requires more than diagnosis. Leaders and workers must believe progress is possible, have enough confidence to invest, and retain the agency to act.
This is disciplined optimism, not blind enthusiasm. It acknowledges that gains and burdens will not be evenly distributed. McKinsey Global Institute estimated in 2023 that activities accounting for up to 30 percent of hours worked in the United States could be automated by 2030, and that an additional 12 million occupational transitions might be needed. Its analysis also found that workers in lower-wage occupations would be much more likely to need to change occupations and that women would face higher transition pressure than men.3 These are estimates, not destiny, but they make one point clear: a productivity strategy without a transition strategy transfers risk to workers.
A productivity strategy without a transition strategy transfers risk to workers.
Optimism becomes credible when institutions pair investment in technology with investment in people. That includes paid learning time, skills-based hiring, internal mobility, apprenticeships, redesigned entry-level work, and partnerships among employers, educators, workforce organizations, and government.
The leadership task is to redesign the system of work
AI strategy is often assigned to a technical team. The technical work matters, but the central decisions are organizational. Someone must decide which problems deserve attention, which outcomes count as value, who bears risk, how work changes, and how people will participate in the transition.
Leaders can begin with five commitments.
Start with the mission and the work. Define the problem before selecting the tool. Map the steps, tasks, handoffs, decisions, and sources of friction in the current workflow.
Design human and machine roles together. Specify what the system may recommend or execute, what a person must review, and who remains accountable. Preserve human judgment where context, rights, safety, cultural understanding, or institutional legitimacy are at stake.
Give workers a role in redesign. The people closest to the work often know where the waste, exceptions, and hidden risks are. Involving them improves implementation and gives them some agency over how their roles change.
Build governance into operations. Treat privacy, bias testing, documentation, monitoring, redress, and incident response as parts of the workflow. A policy that sits outside daily decisions will not govern much.
Measure more than speed. Track quality, worker experience, customer outcomes, error distribution, skill development, and mobility alongside productivity and cost. A system that produces faster output while degrading trust or hollowing out capability may be efficient in the short run and destructive over time. I call this efficiency risk: the danger that measurable near-term gains weaken the human judgment, institutional knowledge, customer trust, or adaptive capacity lowering your competitive advantage over time.
These commitments also redefine management. As AI produces more analysis and recommendations, managers become less valuable as conduits of information and more valuable as orchestrators of strategic intelligence. They must frame better questions, interpret evidence, make tradeoffs, coach people, and take responsibility for decisions. Technical fluency matters, but so do empathy, communication, and ethical judgment. Leadership often requires making decisions that disappoint people, then explaining those decisions in ways people can understand.
Recalibration is continuous
Apply these questions through Human-Centered Agentic Organizational Recalibration™, Dr. Towns’s methodology for human-agent organizational redesign.
There will be no settled endpoint called "the future of work." Each new capability will generate new data, alter expectations, invite political response, and change what institutions and workers believe is possible. The cycle will continue. Leaders therefore need more than an AI implementation plan. They need an institutional capacity for continuous recalibration: a way to sense external change, redesign work, align people and incentives, invest in capability, and learn from the results. In other words they need frameworks for a world in continuous change.
The most important choice is not between humans and machines. It is whether organizations will use AI to narrow work to what can be measured and automated, or to expand what people and institutions can accomplish together. That choice is still ours.
Related Frameworks
Human-Centered Agentic Organizational Recalibration™
Related Research
Sources and further reading
- OECD, OECD Framework for the Classification of AI Systems, OECD Digital Economy Papers No. 323, February 2022.
- White House Office of Science and Technology Policy, Blueprint for an AI Bill of Rights: Making Automated Systems Work for the American People, October 2022. The framework is nonbinding.
- Kweilin Ellingrud et al., “Generative AI and the Future of Work in America”, McKinsey Global Institute, July 2023. The projections are scenario-based estimates published in 2023.
- Edelman, 2026 Edelman Trust Barometer, 2026.
- MIT Sloan Executive Education, Executive Leadership in the Age of Artificial Intelligence, including Beth Stackpole, “Leading the AI-Driven Organization.”
Additional source material: William W. Towns, AI and the Recalibration of Work course syllabus; William W. Towns, The Future of Work: Post-COVID Realities, Policy Changes and the Rise of AI presentation.