Although no one thinks the technology can come true before 2030, companies like General Fusion and Commonwealth Fusion systems are on their way to make it become real.
Robots are now able to educate themselves to deal with the outer world. When it comes to machines taking jobs, industrial robots seem not to be flexible. A robot can pick up an object several times with accuracy on an assemble program. However, moving the object or replacing it with a little difference would be a problem to robot.
However, while a robot is still not programmed to think of how to grasp any object just by taking a look as people can do, it can now get to know how to manipulate the object on itself via a virtual trial and error.
An example of such project is Dactyl, a robot that could educate itself to flip a toy building block using the fingers. Dactyl, which was devised by a company located in San Francisco named Open AI, comes with a robot hand surrounded by lights and cameras. The robot makes use of reinforcement learning, neural-network software in order to learn how to pick and turn the block in a simulated environment. At first, the software experiments in a random manner in order to power connections in the network when it can reach the target.
New nuclear power
New nuclear designs that have been improved over the last several years are expected to make this power source become safer but less expensive. Among them, there are reactors of generation IV, which is an evolution of conventional designs and fusion reactors, which are a technology that seemed to be out of reach. Developers of these designs such as the Terrestrial Energy in Canada and the Terra Power in Washington, have come into the Research and Development partnerships with utilities, with the purpose of grid supply by the end of this year.
Small modular reactors also produce in the tens of megawatts of power. Companies such as NuScale supposed that these miniaturized reactors are able to save much money while reducing risks related to finance and environment.