LogFAQs > #966657819

Whether or not we see it in the future is going to be driven by chemistry and physics more than anything else. We pretty much hit the physical upper limit of the amount of efficiency we can extract from a battery, and those big impressive Boston Dynamics robots require a big ass backpack to do anything for more than 15 minutes. They also have to compromise on things like torque strength on joints in order to keep the robot light enough to not overwhelm the motors, meaning they can't exactly lift very much and they have to do it in very specific ways where they break.

But even if you were resolve the battery problem, the sheer volume of energy you need to generate on a consistent basis just to keep all of them power to produces its own problem, and one that would require a significant increase in something like nuclear power plants to generate that kind of power without completely obliterating the environment.

Once you get all that stuff solved, there's still the issue of onboard processing power. Either your robot needs to have a really strong Wi-Fi connection at all times or you'll need to deal with the fact that the amount of hardware you can fit onto a robot is going to limit how quickly it can make a decision because of efficiency limitations in the algorithms with regard to the chipset. So maybe you can solve the energy crisis, but then you're still stuck with a robot that has to stand at a street corner for 90 seconds before it can make a decision on what to do.

And all of that above is really just about the physical challenges without even touching the broader set of challenges around creating generalized intelligence.