The Next Decade of Work

I have been placing senior leaders for five years. The work has changed significantly in that time, the tools, the candidate pool, the client expectations, the timelines. What has not changed is the core judgment call at the center of every search: whether this specific person, at this specific moment, in this specific company, is the right fit. That judgment has proven stubbornly resistant to automation. I have a view on why.

Every major wave of automation has been accompanied by predictions of mass unemployment that did not materialize in the form predicted. The mechanization of agriculture, the industrial automation of manufacturing, the computerization of office work, all of these displaced specific jobs and created different ones, with a net effect on employment that was far more complex than the simple displacement narrative suggested.

This historical pattern is sometimes cited as evidence that AI will follow the same trajectory: displacement and creation, with the creation ultimately exceeding the displacement as new categories of work emerge. The pattern is genuinely relevant, but it is being used to dismiss concerns that deserve more careful treatment.

What is different about this wave

The automation waves that preceded AI were, with some exceptions, narrow. They automated specific tasks within specific domains, physical manipulation, numerical calculation, information retrieval. They tended to be good at one thing and required human coordination across different things.

Large language models and the broader AI capabilities currently emerging are different in their breadth. The same underlying system can draft legal documents, write code, analyze financial statements, produce marketing copy, answer customer service queries, and synthesize research. The breadth of applicable tasks is not a category shift from previous automation tools, it is a different kind of capability.

This breadth matters because the conventional wisdom about human comparative advantage has relied on complexity and cross-domain coordination as safety zones. If the task requires navigating multiple domains, understanding context across different types of information, and exercising judgment in ambiguous situations, that is where humans were supposed to be irreplaceable. The argument that AI capabilities are now entering significant parts of this space is not obviously wrong, even if the full implications are uncertain.

What actually changes, and what doesn't

The World Economic Forum's Future of Jobs Report 2025 estimated AI will displace 85 million jobs globally by 2030 while creating 97 million new roles, a net gain in aggregate that conceals significant disruption at the individual level.

The jobs most at risk are not the ones that sound the most automated, they are the ones where the core task can be specified precisely enough to train a model on it, regardless of how cognitively complex they appear from the outside. Applying rules to well-defined inputs. Generating first drafts from established templates. Synthesizing information that exists in structured form. These tasks are being automated now, across knowledge work, faster than most forecasters predicted five years ago.

What is not being automated, and what I observe becoming scarcer and more valuable as everything else compresses, is judgment in genuinely novel situations. Not the appearance of judgment. Not pattern-matching dressed up as reasoning. The actual ability to operate in situations that are not fully specified, where the right question is as important as the answer, where accountability cannot be offloaded to a system.

The practical question for anyone building a career or a business in the next decade is not "what can AI not do?", because that list is changing and betting on it is a poor strategy. The more durable question is: what becomes more important as the cost of adjacent tasks falls?

When producing a first draft drops to near-zero, the value of knowing which first drafts are worth developing rises. When information synthesis becomes cheaper, knowing which questions to ask and why rises. When execution of well-specified tasks becomes cheap, the value of specifying the right tasks, and knowing which are worth executing at all, rises. The meta-skill across all of these is judgment: operating effectively in situations that haven't been fully mapped.

What this looks like in practice

From where I sit, placing VP and C-suite leaders, and building infrastructure for hiring operations, this is immediate rather than abstract.

The candidates who are most in demand are not the ones with the deepest functional expertise in a domain that can be automated. They are the ones who exercise judgment in genuinely novel situations, integrate new tools without becoming dependent on them, and build teams that learn faster than the environment changes. Clients have always said they wanted that. The difference now is that the transactional layer underneath has thinned enough that genuine judgment stands out.

The work of finding those people, distinguishing actual judgment from a polished performance of it, has become more important, not less, as AI handles the first pass on everything else. An AI can screen for credentials. It cannot assess whether this specific person, at this specific moment, in this specific company, is the right call. That assessment requires accumulated context and direct observation that no model currently has.

That is the pattern across most of the transition: AI compresses the tasks that felt like work but were actually execution of well-understood patterns. What remains, and what compounds in value, is the human capacity to operate in the space that hasn't been mapped yet.

When the cost of producing a first draft drops to near-zero, the value of judgment about which first drafts are worth developing rises. When information synthesis becomes cheaper, the value of knowing which questions to ask rises. The meta-skill is judgment - operating in situations that are not fully specified.

See also: The Recruiter Isn't Being Replaced. The Job Is Being Redesigned., What Automation Cannot Replace, The Autonomous Hiring Era

Sources

WEF Future of Jobs Report 2025: AI displacement and creation projections

McKinsey Global Institute: Generative AI and the Future of Work in America