For decades, machine intelligence researchers have considered the human brain to
be their target - the thing whose functionality they need to inexpensively duplicate
in order to automate much human labour and compete with human beings in the marketplace.This is a reasonable perspective - but individual human brains are less significant today than they once were. The competition for machines these days mostly consists of networks of humans - and they are competing with them using networks of machines.
This means that cultural evolution is intimately relevant to the work associated with constructing machine intelligence. It is really cumulative cultural evolution which researchers are trying to produce in machines.
This perspective typically leads to a shift of emphasis. Rather than attempting to copy human intelligence, engineers should consider trying to copy human social skills. Today, networks of social machines are proving very useful - even though the individual machines are not very smart - by comparison with the brain of an adult human.
It looks as though ape imitation is very cognitively demanding - requiring the complex ability to put yourself in another person's shoes. However, with machines, transmission of ideas from one machine to another can be done without advanced cognitive faculties - for example by directly copying the relevant area of the machine's brain. It may be easier to produce social machines than smart machines. In other words, a memetics-based approach could well represent a short-cut to producing intelligent networks.
It is sometimes useful to consider the brain as a social network - with the neurons as individual agents and axons and dendrites as their communications network. It's a great proof of the concept that you can make a lot of stupid agents into a larger smart agent - if you have a social network with the correct topology and rules.
This link between cultural evolution and machine intelligence part of why the study of cultural evolution is so important. Not only does cultural evolution underpin most modern human evolution, it is also one of the key subjects for constructing and understanding the intelligent, social machines which will be so important to our future.
Related posts:
- Tyler, Tim (2014) Collective intelligence meets universal Darwinism
- Tyler, Tim (2013) Complexity through socialization
- Tyler, Tim (2013) Memetic algorithms
- Tyler, Tim (2013) Memetic engineering
Posts
No comments:
Post a Comment