Thursday, December 3, 2015

Machines don't evolve?

Just reading an article on the bioethics of CRISPR, and they pointed out that nature--biology--is a vastly complicated mechanism with checks and balances that allow (in an undirected way) responses to many actions.

A mechanical item, such as an old-fashioned watch or a nuclear power plant, has only a limited number of responses to meddling, most of them a little bad (the glass pops off, it runs fast, it runs slow, or it doesn't run) or a lot bad (the nuclear power plant melts down), but "bad" unifies them.

Nature has more responses, most of which we don't notice because the system is so complex. Some are even good (for local values of "good"): this would be evolution in action.

But my sudden thought is, and this thought can be applied to superheroes hence its inclusion on this blog, is that machines don't necessarily have the same mechanism. 

Oh, sapience changes a lot of that if the machines can change themselves (a kind of Lamarckian inheritance, I guess: they make children with the traits that they have found useful), but before sapience, or in the early early stages of sapience, the system is tremendously limited.

There might be an adventure or a story in being the sole intelligent machine and being afraid to mess with your own innards, for fear that you will stop.