Researchers presenting safe way to turn off machine learning




A researcher Googles Deep Mind team has been working with a scientist from Oxford University research presented the possibility to safely disable a self-learning system. They thereby refer to a kind of ‘big red button’.

Artificial intelligence In the paper speaking investigators Laurent Orseau and Stuart Armstrong about the fact that a system based on reinforcement learning do not always learn will show optimal behavior. Therefore, there must be a possibility to turn off such a self-learning system in the event that it exhibits a behavior that can be harmful to the environment or for the person who interacts with the system.

The challenge of the research is to find a way to turn off the system without it finds a way to make “the big red button itself harmless to make it sure that it can not be disabled. This could happen if the system expects certain rewards from harmful actions, so the researchers write. They present therefore a way that deters systems to learn this behavior or ways to seek to avoid elimination.

This falls apart into several pieces, which provide the first part consists sure to be human interruptions not suggest part of the task at hand. By the behavior of the system forced to adapt, it should get the impression chance to have adopted a new policy. Also, the researchers came to the conclusion that certain learning algorithms are easy to interrupt while others require some adjustments for that.

The part of Oxford University involved in the research is the Future of Humanity Institute, which according to Business Insider, among other things focuses on researching threats to mankind. The leader of the institute, Nick Bostrom, is convinced that artificial intelligence within a hundred years is smarter than people and that there is a risk that is against humanity.

This concern is also present at a committee of the European Parliament, so wrote Politico Thursday. The site has been handed a report, which may be cited artificial intelligence surpasses human variation. Therefore, there is research to be done, which addresses the question of how this development into account, for example when it comes to employment.

But safety is one of the topics, as well as the proposal to provide a separate legal status, the most advanced robots, Politico writes according to the report. These robots could acquire the status of an “electronic person” along with certain rights and obligations.


In: A Technology & Gadgets Asked By: [21980 Red Star Level]

Answer this Question

You must be Logged In to post an Answer.

Not a member yet? Sign Up Now »

Star Points Scale

Earn points for Asking and Answering Questions!

Grey Sta Levelr [1 - 25 Grey Star Level]
Green Star Level [26 - 50 Green Star Level]
Blue Star Level [51 - 500 Blue Star Level]
Orange Star Level [501 - 5000 Orange Star Level]
Red Star Level [5001 - 25000 Red Star Level]
Black Star Level [25001+ Black Star Level]