Have you ever wondered what would happen if we reached a
point in which robots carried out all of the mundane tasks that we don’t want
to do? Then what if these robots learned to optimize these tasks and update
themselves based on real world experiences? Then what if they stopped updating
to do the right thing, but instead to do the wrong thing? What if this led to robots
plotting against humans in an attempt to takeover the world? Yes, what I just
explained is the plot from “I, Robot” but the general line of thinking is more
realistic than you would think.
 |
| AI Robots in "I, Robot" |
Google researchers have recently developed an AI kill switch
in order to shut down an AI robot no matter what. This is incredibly important
because AI robots often work by trying to maximize some sort of function or
algorithm. They will then read real world events as inputs and adapt those
inputs into their systems so that they can continue to maximize their
functions. Human interference can be considered one of these real world events.
Therefore, if the machine determines that human interaction is detrimental to
maximizing their functions or that the human is trying to shut it down, the machine could act out against the
human to stop this from happening. Due to this fear, Google created an AI kill
switch to automatically shut down a robot at anytime.
However, Google would never want to use the kill switch unless it had to and
is instead working on creating a way to have robots read human
interactions as part of the task and therefore not as a detrimental
interruption. Doing so would essentially allow humans to teach robots things
such as to not go outside in the rain. Hopefully, through both teaching robots
how to read human interactions and including a kill switch in the machines, we will
never have to live through the reality in “I, Robot.”
 |
Resources:
- http://motherboard.vice.com/read/google-researchers-have-come-up-with-an-ai-kill-switch
- http://www.themanufacturer.com/wp-content/uploads/2015/08/EPA245.jpg
- https://www.singularityweblog.com/wp-content/uploads/2015/02/Kill-Switch.jpg
- http://images.contentful.com/7h71s48744nc/3l9IwugJKgcCq8YMwWGiy/8fdbc22b56ca7a2d34fb26e76952533e/i-robot.jpg
Hi, I like your article, it is so interesting. I don't even know that robots can read, absorb real world events to maximize their functions by themselves before reading your article. It is an awesome technology improvement but kinda terrific. I can't image what will happen if robots have their own thoughts and want to dominate the world. It's lucky that we have a lot of smart people to solve this problem. I believe that the AI kill switch will definitely control the robots somehow. Nice article.
ReplyDeleteDoes the Al Kill switch make the robot become not usable anymore? It sounds like it destroys the robot. I would assume that since humans are the ones that create and build robots they're responsible for making sure that the algorithm has no cracks and works well. There should be no need for a kill switch because the robots should have no reason to hurt humans. Regardless awesome blog and interesting topic.
ReplyDeleteIts crazy that we have progressed so far to the point where robots could be able to read human interactions. Although I think robots that mimic humans are incredibly cool, I am definitely freaked out by the thought of a robot population growing one by one and taking over the world. The Al kill switch is a good safety measure, I just hope the robots don't find the switch, destroy it, and then destroy the world!
ReplyDeleteWow that's pretty crazy. It's the kind of thing that you wouldn't expect programmers to actually have to plan for, but I guess that shows how far along they've come with making autonomous robots. Do you know how widespread the implementation of this technology is, as in is it being applied to all robots or more high powered ones?
ReplyDeleteI think thats pretty cool. Its weird to think that you would ever want to cause your own program to stop working. After all you work so hard just to get your program up and running. At this same time it is cool that google is thinking about all potential scenarious and planning for them.
ReplyDeleteA real cool subject.I really didn't know that you could program a super computer or a robot to do something as crazy as that but that raises the question of is that safe for us as a population that might be kicked out really quickly with such advanced technologies
ReplyDelete