Will AI Work For Us, Or Will We Work For AI?

Initially published on Forbes October 22, 2026

Imagine a centaur, the creature from Greek mythology with the body and legs of a horse and the upper body of a human. The human looks ahead, chooses a direction and draws on the horse’s strength to get there.

Something like this is taking shape as we combine human and artificial intelligence at work. But the way we put humans and AI together matters. Is artificial intelligence the powerful body, with a human head directing it? Or are humans becoming the body, carrying out instructions from an artificial intelligence that supplies the direction?

What kind of world are we building?

Speaking at UNLEASH 2026 in Paris, Peter Hinssen, co-founder and partner at nexxworks, challenged a room full of HR leaders to examine that choice. Drawing on Cory Doctorow’s reverse-centaur metaphor, he invited them to consider which jobs would give people greater capability and which would turn people into executors of an algorithm.

When AI At Work Weakens Human Agency

The answer depends on how organizations see people at work. If their value lies primarily in carrying out assigned tasks, technology can replace much of what they do today. Hinssen made this visible through warehouse work: the algorithm determines the action, and the person repeatedly performs it. In knowledge work, the arrangement can be harder to recognize. A person may still review the recommendation and approve the decision while contributing less of the thinking that shapes it.

Wharton researchers Steven Shaw and Gideon Nave call this “cognitive surrender”: accepting AI-generated answers with little scrutiny, allowing the system’s reasoning to substitute for your own. Their 2026 working paper describes three experiments involving more than 1,300 participants answering reasoning questions, with or without AI assistance. Participants could follow incorrect AI advice while becoming more confident in their answers. That means a human can remain responsible for the answer while signing off on what AI recommends without questioning it.

There is another way human judgment can disappear from the work. At UNLEASH, University of Bath Professor of Practice Margaret Heffernan described “organizational silence”: people have ideas and concerns but withhold them because they fear consequences or believe nothing will change. They may see a different direction and have learned that there is little point in proposing it.

Cognitive surrender and organizational silence are different, but they expose a shared weakness in the promise of human-led AI. People can stop forming an independent view, or stop contributing the view they have. Either way, an organization can continue to employ their minds while drawing on less of what those minds could offer.

That is how the reverse centaur can take shape in knowledge work, even while the process is still described as human-led.

Building a different future means expanding the human capacity to decide what is worth accomplishing alongside the technology’s power to accomplish it.

Human-Led AI Starts With Human Judgment

Start a transformation conversation with the technology, and the questions follow naturally. What can it automate? How quickly can it produce? Which activities can it absorb? How much could we save?

These questions help organizations understand a new capability. They can also make existing work the boundary of future ambition. We look at what we already produce and imagine producing more of it, faster and with fewer resources.

Human creativity offers another starting point. Someone recognizes a need we haven’t addressed, imagines an experience that doesn’t yet exist or questions an outcome everyone else has accepted. Technology then gives that intention greater reach.

Consider the difference between using AI at work to handle more customer complaints and using what people learn from those complaints to reconsider the service itself. The first objective is to improve an existing process. The second might lead someone to recognize that the process is organized around a problem we should be trying to remove.

AI can help with both. But it takes a human to decide which way to go.

The head of the centaur represents that responsibility: choosing what deserves our effort and what a better outcome would mean.

Human Creativity Needs Development Too

It is tempting to assume that this direction will come naturally. Let machines handle execution, and people will move into creativity, judgment and meaningful work.

But releasing people from tasks doesn’t automatically develop those capabilities. Developing creativity and judgment deserves as much attention as learning to use the tools.

Heffernan argues that creativity is a capacity people already possess and organizations need to develop and support. She draws on the example of artists who pay close attention to the world, collecting observations before knowing what might emerge from them. They begin without a complete plan and develop the work through choices and revisions. Their direction emerges through that engagement.

These capabilities matter when people need to recognize possibilities before they can define the task. Creating a different service, a more humane system or a new strategy requires people to originate work they were never assigned. Before there is a prompt, someone has to notice something worth pursuing.

That means giving people opportunities to notice, investigate and create — and giving their discoveries a route into decisions. An employee’s observation can become a team’s experiment and eventually a change in direction. AI can help more people turn an idea into something others can examine and use. Leadership determines whether those ideas have somewhere to go.

AI Transformation Needs A Bigger Definition Of Progress

Efficiency matters. Removing unnecessary effort can make work better, services more accessible and new possibilities affordable. The problem comes when efficiency becomes sufficient evidence that we have built a better future.

The reverse centaur could perform extremely well against that measure.

We need a fuller ambition for the combination of human and artificial intelligence. Alongside what it produces, we should examine what it enables: more people able to pursue an idea, solve a problem they care about, contribute their experience or change a system that isn’t working. That opportunity needs to extend beyond the leadership team. People throughout the organization encounter needs and possibilities that those setting the strategy may never see.

The centaur gives us an image of power directed by human intention. Building it requires us to take both halves seriously. We are investing heavily in what artificial intelligence can become. The world it helps create will also depend on the human imagination we develop, and how many people get to use it to choose the direction.