Robo-Logic

Interesting video. Robots “learn,” but within a very narrow set of parameters. Those robots featured are created for a very specific task, one that requires the accurate repetition and endurance that humans can’t consistently delivery. So they don’t learn so much as practice their programming until a specific standard of precision has been met.

Learning, the way humans do, requires a kind of meta-logic—probably many layers of it—that humans develop as a matter of course. We may know how to perform function x reasonably well, but how do we know when to perform one function x, as opposed to function y? We just do. We can consider, consciously and automatically, many different inputs that robots are unable to attend to, especially abstract inputs like linguistic or social cues. The computing power necessary for robots to learn like this is nowhere near as strong or complex enough, not to mention how much power is needed to support such a system. The industry is a long way off from that level of product viability.