Technology

Google Developing Computer That Would Program Itself, Think Like Human

| by Michael Allen

DeepMind Technologies, a company owned by Google, is working on the Neural Turing Machine (NTM), a computer that would ultimately program itself.

According BetaBeat.com, the goal for the NTM is to create computer programs for unfamiliar situations just as a humans do today.

The NTM uses "neural networks," which have been around for years, but takes this artificial intelligence to a new level.

Popular Video

People were so furious about this Pepsi ad that Pepsi pulled it after just one day. Watch it here and decide if it's offensive:

"These neural networks that are so good at recognizing patterns, a traditional domain for humans, are not so good at doing the stuff your calculator has done for a long time," Jurgen Schmidhuber of the Dalle Molle Institute for Artificial Intelligence Research in Manno, Switzerland, told New Scientist.

If successful, the NTM would have the calculating power of a normal computer combined with adaptability of a human, enabling it to write programs for new situations.

Popular Video

People were so furious about this Pepsi ad that Pepsi pulled it after just one day. Watch it here and decide if it's offensive:

During a recent test, DeepMind Technologies found that the NTM learned faster and could reproduce longer blocks of code than a normal neural network.

The NTM's methodology resembled a human programmer.

"As humans we classify but we also manipulate the classification," Chris Eliasmith of the University of Waterloo, Canada, told New Scientist. "If you want to build a computer that is cognitive in the way that we are, it is going to require this kind of control."

"Digital computers are basically hitting a wall," added Eliasmith. "If we want to use our most efficient hardware, we have to express our algorithms in a manner which fits."

Another team from Google recently announced that it created a neural network that can reads and executes simple computer code without being taught the programming language.

Sources: BetaBeat.com, New Scientist (Image Credit: Gengiskanhg)