There is a lot of noise about AI destroying jobs. That concern is mostly overstated.

What we are seeing in practice — and what much of the better research and operator commentary points to — looks different. Most AI systems still need a human directing them. Even with agentic AI, someone has to set the goal, provide context, judge the output and own the result.

AI is already changing the composition of work. It takes over chunks of execution. It does not remove the need for judgment, taste, accountability or decision-making.

Used well, it makes people dramatically more productive: the same person can produce more, improve quality and spend less time on the mechanical parts of the job.

Jobs are bundles of tasks

The mistake in the "AI will make everyone redundant" narrative is treating a job as a single unit that either survives or disappears.

Jobs are bundles of tasks. AI is excellent at some of those tasks. It is much weaker at the full loop:

  • Deciding what matters
  • Navigating ambiguity
  • Weighing trade-offs
  • Taking responsibility for the outcome

That is why, in most real use cases, the winning setup is not "AI instead of people." It is "people directing AI."

There are nuances by role. Some work is more automatable than other work. Some companies will use AI as an excuse to cut. That does not make mass replacement the default path.

The ceiling on individual output moved

The larger shift is that the ceiling on what one capable person can produce has moved up.

Companies that understand this will use the leverage to do more, not simply to do the same with fewer people. AI is not mainly a job destroyer. It is a productivity multiplier for people willing to learn how to use it.

The people who remain effective will be the ones who can:

  • Direct AI clearly
  • Evaluate its output critically
  • Redesign work around what humans should own and what agents should execute
  • Keep learning as the tools improve

The upside is that AI also makes that learning easier. The tools to level up have never been more accessible.

So the more useful leadership question is not simply whether AI replaces people. It is whether people will choose to adapt — and whether leaders will redesign work so humans and agents compound each other.

In most of the work we are seeing, the human stays in the loop. That human becomes considerably more powerful.