If you have watched the World Humanoid Games in China, you may have seen something more revealing than a spectacle of machines. Beneath the karate kicks, somersaults and backflips lies a glimpse of where we are — and perhaps a reminder of just how far we still have to go.

The mechanics of building a bipedal humanoid robot are advancing at extraordinary speed. Robots can now balance, walk, run and manipulate objects with a dexterity that would have seemed almost impossible less than a decade ago.

China is pushing this progress rapidly, producing machines capable of astonishing athletic feats. But behind the spectacle is an important truth: much of what we see is carefully programmed, rehearsed and controlled. The robots can perform remarkable movements, but performing a movement is not the same as understanding it.

And this is where the Humanoid Robot Games become particularly interesting.

They reveal that the next great challenge may not be the robot’s body, but its mind — or, more precisely, its computing power. A human brain performs an extraordinary number of calculations while simultaneously balancing, seeing, predicting, adapting and making decisions, usually without us even thinking about it.

For a machine, even apparently simple human behaviour can require enormous computational effort.

That is perhaps the most fascinating lesson of all. The closer we get to building machines that resemble us, the more extraordinary human beings begin to look.

We walk without calculating how to balance. We reach for a glass without consciously solving a physics equation. We recover from a stumble before we have even realised we are falling. What feels effortless to us can be extraordinarily difficult for a machine.

Greater computing power will undoubtedly push robots into a new realm of capability. But when that happens — and how quickly — remains uncertain. Technological progress is rarely a straight line.

For now, the Humanoid Robot Games demonstrate something both exciting and humbling: we have made phenomenal progress, but we are still some distance from machines truly replacing humans.

I said not yet.

And perhaps the most interesting question is not whether robots will eventually become like us, but what we will discover about ourselves while trying to make them so.