The Future of Work Starts With the Work Itself

The visible future of work has already arrived

For years, the “future of work” was discussed as if it were a destination somewhere ahead: more remote working, more automation, more freelancers, different generations and an endless requirement to reskill.

That framing has become less useful.

Many of the visible elements of that future have already arrived. In the United States, 45.2 per cent of employed adults reported using generative AI for work in the second quarter of 2026. Among US employees whose jobs can be performed remotely, Gallup found that 52 per cent were working hybrid in May 2026; another 26 per cent were fully remote.

These figures describe specific US populations, not the global workforce. But they make one point clear: AI use and flexible work are no longer edge cases waiting for executive permission.

The harder problem begins after adoption.

An organisation can introduce copilots, create a hybrid-work policy, launch a skills academy and still leave the work itself almost untouched. The same jobs exist. The same approvals slow decisions. Managers still control information. Learning remains separate from actual work. Employees acquire new tools but continue executing processes designed for a different technological era.

The future of work is no longer primarily about where people work or how many tasks AI can automate. It is about how work is decomposed, distributed and made accountable.

Stop treating the job as the smallest unit of work

The public AI debate still gravitates towards occupations: Which jobs will disappear? Which professions are safe? What will be the careers of the future?

But organisations do not automate occupations. They automate, augment, combine and redistribute tasks.

That distinction matters. The International Labour Organization’s 2025 global assessment found that one in four workers was employed in an occupation with some degree of exposure to generative AI. Yet because occupations contain many different tasks and still require human input, the ILO concluded that transformation of jobs is the more likely overall effect than wholesale replacement.

Ravin Jesuthasan and John Boudreau have argued for years that traditional jobs should be deconstructed into tasks and activities, with work then reconstructed around the capabilities required to perform them. AI makes that idea more operationally urgent.

If one part of an analyst’s role can be automated, another accelerated by AI and a third depends on contextual judgement or stakeholder trust, asking whether “the analyst job” will be automated is the wrong question.

The useful questions are more granular. What problem is this work solving? Which tasks create value? Which can now be automated reliably? Where can AI increase the capability of a human expert? Where would automation introduce unacceptable risk? And where does human judgement matter precisely because the answer cannot be reduced to a repeatable rule?

Once leaders begin there, work design becomes strategic rather than administrative. It also changes talent strategy. Instead of asking only who holds a job, organisations can ask where the required capabilities exist and how they can move towards the work.

But it creates a harder management problem: accountability has to be reconstructed alongside the work. Someone still has to own the outcome.

AI productivity is real. Organisational productivity is harder.

There is credible evidence that generative AI can improve performance on particular tasks.

A field study by Erik Brynjolfsson, Danielle Li and Lindsey Raymond, published in The Quarterly Journal of Economics in 2025, found a productivity increase of roughly 14 per cent among customer-support agents using an AI assistant. Stanford’s 2025 AI Index reviewed a growing body of workplace studies showing meaningful productivity improvements across several categories of work.

Yet this creates an important trap.

A task becoming 20 per cent faster does not mean the organisation becomes 20 per cent more productive.

If the resulting work then waits three days for approval, creates more output for another overloaded department, increases verification requirements or accelerates a process customers never valued, the local efficiency gain can disappear inside the system.

Ethan Mollick’s work offers a useful way to think about the problem. Research with Boston Consulting Group described a “jagged technological frontier”: AI performs extremely well on some apparently sophisticated tasks yet can perform badly on others that look deceptively similar. Workers therefore need more than access to AI. They need the judgement to understand when to delegate, when to collaborate and when to distrust the machine.

The implication is bigger than prompt training. Companies have to redesign workflows around that frontier.

That means specifying where AI enters the process, what information it may access, which outputs require review, who has decision authority and what happens when human and machine assessments conflict. As AI moves from assisting with individual tasks towards executing sequences of work, those choices become part of everyday management.

The differentiator is not how many AI tools an organisation deploys. It is how quickly the organisation can redesign work as machine capabilities change.

Hybrid work is a coordination architecture

Something similar has happened with hybrid work.

For remote-capable work, years of argument about whether people should be “in the office” or “at home” have obscured a more useful question: What type of interaction does this work actually require?

Gallup’s US data suggests that hybrid work has stabilised rather than disappeared. That does not prove that hybrid is optimal for every organisation. Manufacturing, healthcare, hospitality and many operational roles have fundamentally different constraints. Nor does it tell a CEO whether Tuesday and Wednesday should be mandatory office days.

What it does show is that work location has become an enduring design variable.

Lynda Gratton’s work on redesigning work argues against treating flexibility as a universal perk. Different tasks benefit differently from focus, collaboration, energy, coordination and connection. The objective is therefore not maximum flexibility or maximum attendance, but an intentional configuration of time and place around what helps people perform.

Tsedal Neeley adds another layer: distributed work depends on deliberately creating trust, communication norms and social connection rather than assuming proximity will create them automatically. Her research on global and virtual teams predates the pandemic and places the management problem squarely in coordination and culture.

Seen this way, return-to-office mandates and remote-work policies are both incomplete answers.

The real questions concern which activities need synchronous interaction, what merits physical presence, how knowledge is documented, how decisions remain visible to distributed colleagues and whether performance is assessed through outcomes rather than proximity.

The office becomes one component of work design rather than the organising principle of work.

Skills only matter when they move with the work

The same problem appears in reskilling.

The World Economic Forum’s Future of Jobs Report 2025, based on responses from more than 1,000 large employers across 55 economies, found that employers expected 39 per cent of workers’ core skills to change by 2030. AI and big data were among the fastest-growing skill categories, but so were human capabilities including creative thinking, resilience, curiosity and leadership.

The usual organisational response is more training. But a catalogue of courses does not create a skills-based organisation.

Skills become strategically useful when leaders know which capabilities the redesigned work requires, can see where those capabilities exist and can move people towards opportunities to apply them. Learning then becomes part of the flow of work rather than a parallel HR activity.

This also changes the manager’s job.

Deloitte’s 2025 Human Capital Trends research found a striking execution gap: 73 per cent of organisations surveyed recognised the importance of reinventing the role of the manager, but only 7 per cent reported making great progress.

That gap matters because redesigned work redistributes authority. Managers accustomed to supervising jobs and employees may instead need to orchestrate capabilities, allocate work, develop people, govern AI-supported decisions and create clarity across teams that are less dependent on hierarchy.

The future manager is therefore not simply a more empathetic version of the old manager. The role itself is being redesigned.

Four questions reveal whether the work is really changing

Leaders do not need a definitive blueprint for the future of work. Machine capability, labour markets and organisational expectations are moving too quickly for that.

They do need a disciplined way of examining work as it changes.

The answers will differ between an investment bank, a hospital, an industrial manufacturer and a technology company. That is precisely the point.

There is no single future of work. There are different forms of work being rebuilt under new technological and human constraints.

Organisations that understand this will stop treating AI, hybrid work, skills and culture as separate transformation programmes. They are different dimensions of the same design problem: how to organise work so that talent, technology and decision-making reinforce rather than obstruct one another.

That is where the future of work becomes less about predicting what comes next — and more about designing an organisation capable of changing when it does.

Four lenses on the same work-design challenge

“Future of work” can mean radically different things to different audiences. The right keynote depends on the work-design question the room needs to confront.

Work deconstruction

Ravin Jesuthasan

How to break jobs into tasks and rebuild work around the capabilities of people and machines. ↗

Human–AI collaboration

Ethan Mollick

How leaders and knowledge workers navigate AI’s uneven and rapidly changing capability frontier. ↗

Intentional work design

Lynda Gratton

How time, place, energy and human needs can be configured around the work rather than blanket policies. ↗

Distributed coordination

Tsedal Neeley

How trust, communication and culture are built when teams operate across locations, technologies and boundaries. ↗

Those are related conversations, but they are not interchangeable. For adjacent perspectives, explore Scurati’s work on Digital Transformation, Change Management and Leadership & Corporate Culture.