At the University of
Washington there is a group of computer scientists and psychologists who are
working together to develop a truly learning robot. Currently in order to “teach” a robot to do
something it has to be specifically programmed in that it should do that
task. This is significantly different
than how humans learn and because we all like to compare everything to humans
this group is attempting to create learning robots. Human babies learn through observations; they
see what their parents and other people whom they interact with do and imitate
them in order to learn new skills. The
idea of a truly learning robot that could learn by observing other people would
be very significant and would be a huge step in robotics. By making this advancement in robotics the
area of human robot interactions would be greatly changed and a whole new era
of robotics would arise.
The creation of
learning robots would bring new benefits to the field of robotics as well as
the world as a whole. By having robots
that learn there would be increases in productivity in workplaces that rely on
robots to perform various tasks around the workplace. It could also improve the safety of many
tasks as it can be hard to program a robot to do certain things that can be
dangerous tasks. One example of this could
be things such as firefighting and search and rescue missions.
There are also obvious
negative consequences that would come from creating robots that can learn. By creating learning robots many people could
lose their jobs. Many simpler jobs would
most likely be replaced. On example of
this are people who place items on conveyor belts to go through a robotic
assembly line. If robots could learn
these people could easily be replaced as the issue of not having a precise spot
to grab the initial item from prior to its placement on the assembly line would
be eliminated. The robots would learn to
be able to find where the item they needed to grab was so not being able to
program a specific location to grab from would no longer be a barrier that
prevents robots from taking this job over.
Another possible issue is that we won’t know how much these robots would
be able to learn and how far it could go.
We wouldn’t want it to result in a Terminator like scenario in which the
robots decide to over through the human population and kill us all off.
Obviously much of this
is speculation and it very likely might not happen. The project is still in its very beginning stages
and only two scenarios have been observed with the project. The idea is also to teach the robots to learn
like babies and young children so the level at which the robots are anticipated
to learn at is very low. Although this
is the plan it is still important to think about how far this could possibly go
and prevent the robots from being too smart.
http://www.popsci.com/robots-could-learn-same-way-babies-do
http://journals.plos.org/plosone/article?id=10.1371/journal.pone.0141965
No comments:
Post a Comment