Discussions about the future of work often coalesce around one major point of contention: the impact of automation on the workforce. Pessimists believe that humans will be made redundant by artificial intelligence (AI) and robots, leaving them unable to find work in a future bereft of jobs. Optimists believe that historical norms will reassert themselves and technology will create more jobs than it destroys, resulting in new occupations that require new skills and knowledge and new ways of working.

If work is viewed essentially as a collection of tasks, then AI’s growing capabilities may indeed seem troublesome.

Rarely does anyone engaged in this debate step back to examine what is meant by “work” itself. Yet both the pessimistic and optimistic views are founded on a culturally bound conception of work, shaped by the ideas and practices of the Industrial Revolution. In this conception, work is seen as the performance of a well-defined task or set of tasks, laid out sequentially end to end, in assembly-line fashion, to achieve a particular outcome. Efficiency gains come from specialization, which allows workers to become better and faster at a given task through practice, and from automation, which replaces the human task performer with an even better and faster machine.

If work is viewed essentially as a collection of tasks, then AI’s growing capabilities may indeed seem troublesome, raising the specter that most or all human work will simply be automated away. But is it time, in this post-industrial age, to consider a different path? As AI becomes more capable and flexible, might it not enable work itself to be reconstructed—not as a set of discrete tasks in a process, but as a collaborative problem-solving effort in which humans define the problems, machines help find the solutions, and humans verify the acceptability of those solutions?

Atomizing work into a predefined set of tasks suits neither human nor intelligent machine. To be sure, people can perform specialized tasks, and AI can be used to automate them. But realizing our full potential—and that of our technologies—may lie in putting them both to a more substantive use, with each augmenting the other’s capabilities.[1]

Consider how humans and machines could productively interact if work were organized around problems to be solved, not processes to be executed. In such an environment, management of the problem definition becomes the main concern.[2] Humans take responsibility for shaping the problem—what data to consider, what good looks like—and for evaluating the appropriateness and completeness of the solution. Automation, including AI, augments this work with a set of digital behaviors[3] that replicate specific human actions—but with the advantage of using more data to provide more precise answers, while not falling prey to the cognitive biases to which humans are prone.

Reframing work from task to be done to problem to be solved—and the consequent reframing of automation from the replication of tasks to the replication of behaviors—could give us the opportunity to make the most of AI’s capabilities, as well as our own.

For more insights on Future of Work visit:

Deloitte Insights


[1] Jim Guszcza, Harvey Lewis, and Peter Evans-Greenwood, “Cognitive collaboration: Why humans and computers think better together,” Deloitte Review 20, January 23, 2017.

[2] We should note here that shifting our focus from process to problem enables us to make processes malleable, rather than being static. AI technologies already exist—and are, in fact, quite old—that enable us to assemble a process incrementally, in real time, allowing us to more efficiently adapt to circumstances as they change. This effectively hands responsibility for defining and creating processes over to the robots—yet another complex skill is consumed by automation.

[3] We note that behaviors are not necessarily implemented with AI technologies. Any digital (or, indeed, non-digital) technology can be used.