How Smart Should We Allow Robots to Get

Do Now

How smart should we allow robots to get?

Introduction

Robots that can heal themselves, big-rig trucks that take over driving for tired drivers, phones that answer your questions, thermostats that know your preferred temperature – artificial intelligence is now available for more tech gadgets than ever before, and it is getting smarter every year. While some scientists envision artificially intelligent robots as rescuers and caretakers of the future, others are worried that creating technology as intelligent as humans might be rather dangerous.

Artificial intelligence (AI) is intelligence shown by devices, machines, or software. AI can be as simple as getting a computer to “act” intelligent by creating software that allows the computer to perform an intelligent behavior, such as transcribing speech into text or solving a Sudoku puzzle. Even Apple Siri and Google More advanced AI might incorporate multiple intelligent behaviors in one device, use sensors or human input to make decisions, or grow more intelligent by learning from past experiences.

“Robots are going to provide tremendous benefits to society one day, especially if they can do tasks that are too dangerous for humans,” said Jeff Clune an assistant professor at the University of Wyoming in Laramie, in an interview with Science Friday. Clune is also the director of the Evolving Artificial Intelligence Lab, and recently co-authored a research paper unveiling a six-legged robot that is able to use previous experiences to find a new way to walk after a leg injury.

“Eventually we want robots to put out forest fires and find survivors after earthquakes. But we won’t be able to realize those benefits if a robot gives up at the first sign of problems. What we need instead is robots that can soldier on if they have some damage to keep doing the mission that we’ve asked them to do,” Clune said.

Though using AI to create a self-healing robot might sound a little creepy, there are many ways that AI has subtly crept into our everyday lives. AI is already being used to predict flu outbreaks, make internet advertising more effective, drive trucks and operate warehouse machinery, and is increasingly a part of common household appliances like thermostats and refrigerators. For now, AI like iPhone’s Siri are seen as convenient tools, but what happens when AI outsmarts its creators?

“The problem with really, really smart machines, is that if you don’t give them the right instructions, if you don’t give them objectives that are perfectly aligned with what the human race wants, then you have a problem,” said Stuart Russell, professor of computer science and engineering at UC Berkeley in an interview with Science Friday. “If the machine is much smarter than you, it means that it can take more information into account, it can look further ahead in the future, [and] it can anticipate all of your counter moves. Now you are kind of in a losing situation.”

So how smart do you think we should we allow artificial intelligent to become?

Source : kqed