[dba-Tech] Robotic cats, etc.

[Arthur Fuller]

One aspect of robotics that completely escapes me is the need to make them
similar to mammals and arachnopods (not insects, since they have 8 legs not
6); i.e. the presumption that such bots will need legs capable of climbing
a staircase, and arms capable of opening a can of pork&beans using either
their very sharp nails or perhaps a convention can-opener.
I want to question the central premise of such experiments. Why should bots
bear any resemblance to us bipeds or dogs or cats or spiders? Why not think
way outside the margins instead? One possible answer is that evolution has
demonstrated that these architectures work: cheetahs can run faster than
any other mammal, eagles can spot a mouse from 300 feet, humans can
identify minuscule genetic differences and justify the murder of millions
of people (where do you want to start that discussion? South Africa,
Rwanda, Korea, China, and let's not forget the Americas, where entire
civilisations were murdered.) I digress, sometimes I can't help it. Call it
a political form of ADHD. Does training RAI (a term I just coined which
stands for Robotic Artificial Intelligence) to emulate human behaviour mean
that these machines will learn from us how to murder millions of people,
take all their assets and hide them in Swiss and offshore money laundries?
Is that what we want robots to learn?
IMO, the Roomba is the way of the future. It looks nothing like any animal
you can think of, but it is smart, albeit limited in its functions, but
that's my point. A robot with AI built in doesn't have to resemble any
existing spider or insect or mammal; in addition, it has to learn from
previous experience, as in the case of IBM's Watson.
I plan to extend this discussion but I'm also cautious of word-length,
which I may have already violated. So, stop for now. More to come.

-- 
Arthur