Google is on its way to making the world a much better place through artificial intelligence. Google’s DeepMind AI has seen another improvement through the integration of the Differential Neural Computer or DNC. DNC is actually a memory-augmented neural network which uses its stored memory to answer questions about complex, structured data.
In simpler terms, DNC uses this external memory together with the neural network in order to solve a problem. Now, making this possible is a controller that sits in the heart of a DNC. According to DeepMind researchers, this controller is comparable to a computer’s processor. Taking input in, reading from and writing to memory as well producing output that can be interpreted as an answer are the tasks of the controller.
The new capability of DeepMind’s artificial intelligence was manifested in the two tests conducted. First, the AI was fed the basic information about the London Underground public transport system and then was asked to find the shortest route to certain points. Amazingly, it was able to complete the task with 98.8 percent accuracy. Second, the AI was given information about the relationships in a family tree and was able to come up with relationships even though it wasn’t fed with it.
Now, we only used to see artificial intelligence in movies but, with this latest development, it won’t be surprising if AI becomes mainstream in a few years.