I came across an article on a proof-of-concept for a cloud-based knowledge bank for robots. The basic idea is that physical robots can be connected to a centralized collection of data and algorithms that can augment the robot’s capacities and, at the same time, collect learning from the robot for other robots to use.
The fact that the robots share a common, expanding knowledge base is of particular interest in light of artificial intelligences that play a big part in Sea of Dreams. I like that, in some ways, the robots are both distinct and part of ‘one mind’. Their experiences could become the possibilities for others to do something different in a different context.
Say, for example, one robot has an experience of navigating a lunar landscape. That experience is stored as a code to manage its wheels and drive shaft in order to navigate the moonscape more effectively. Suppose then another robot were to use these same instructions to move across a desert looking for oil. It seems to me that some sort of creativity has occurred: applying the experience of one field to another.
This also raises the question of how the robot chooses what data and algorithms to use. Does it have a way to look at the moon rover data and say, “That situation is similar to the one we are in, so we can use it.”