According to this recent article, researchers have started to overcome the last challenge to unlimited Artificial Intelligence. They succeded in creating a very simple robot and use an evolutionary algorithm so that the robot can learn to walk (or jump, as the case may be) just like a human child learns. But then they added a little bit of magic:
MacLeod’s team took this idea a step further, however, and developed an incremental evolutionary algorithm (IEA) capable of adding new parts to its robot brain over time.
What this means is that when the robot’s performance and learning has stopped, then another system takes over and modifies the robot’s brain. This is an incredibly powerful and promising approach, and the challenge in the past has been limited computing power. The reason is that you need an exponentially faster processor to carry out this brain modification functionality.
Fortunately for us, accelerating technology will remove this challenge. Right now computer power is doubling approximately every year. The result is that in just three or four years we can buy a computer 10 times as powerful for the same money. The really exciting part is that it will make it possible to create artificial intelligence of unlimited power.
The researchers are working on creating a simple robot that moves along the tabletop. However, they will be able to apply the systems that control brain modification to all future robots that they create. It’s only a matter of time before this technology is applied to a more complex system such as vision recognition or stock trading. This will create a system that actually works better than the human counterpart.
Many of these ideas may sound silly today, but just 20 years ago our current level of technology could be imagined by only a few people. Anyone who is aware of the power of accelerating technology can visualize Artificial Intelligence systems that are more powerful and work much better than their human counterparts.