A Billion Humanoid Bots By 2040?

Expanding on the comment from one of previous articles:

A Billion Humanoid Bots By 2040?

Not a chance. Not because it can't be done, but because humanoid robots are a bad solution to the problem they are supposed to solve.

A fully autonomous vehicle is great, putting a humanoid robot behind the wheel of a regular car, while it might work with sufficiently advanced robot, makes zero sense. The same is true for most of the tasks. How can that, even after years of improvement, ever be more efficient than those? A lot of tasks can't even be performed by humanoids due to the shape. F.e. pipe inspection. You need a soft robot that can learn how the pipe made of particular material should look like, how can it look after years of exposure to the substance it is used to carry, what type of damage or deposits might have built up inside, then get into the pipe, navigate it with no preloaded map (because documentation might be incomplete), recognize problems and carry on scraping sludge, pulling out clogs, filling cracks and patching leaks, all without control or even mere communication with people (because pulling cable might cause problems and wireless might have limited range). So learning, navigation and adaptability? Yes, yes, yes. Humanoid? No. Universal? No. Planes are viable vehicles because they do not copy the way of flying of any animals. Only recently people started to make drones that fly like birds, bats or insects, and mostly for fun / as proof of concept. The same way humanoid robots are a dead end as a product that can perform work, and in the same time a pinnacle of robot engineering as they are that hard to make.

Big tech developing humanoid robots makes sense, because by the time they are made viable, engineers will have to solve several problems and that experience is going to be used to improve robots that are actually doing the tasks efficiently. The main point: there is no harsher environment for the robot than the one where it has to perform its tasks along humans.

There will be billions of robots that can map and navigate the environment on their own, learn the same way humans learn (but faster) and adapt to specific tasks with no supervision, but only a tiny fraction of them will be humanoid.

Certainly there will be specialized robots. Hell, we already have that with Roomba to a degree. However, I think the point you are missing is we are referring to general purpose robots where the system has the knowledge of all kinds of jobs.

For example, lets use domestic robots. If you take all the tasks associated with a home, you could specialize some of them but, if you wanted a general purpose that could handle it all, what better form than a human. Surely there will be adaptations with certain things removed with others added.

Under your scenario, depending upon what one's precise definition of a robot, we could be looking at higher numbers if we include specialized robots into the equation.

We will see how the development proceeds the next 10 years.