Google DeepMind AI Can Now Learn on its Own

DeepMind, the artificial intelligence (AI) developed by Google parent company Alphabet, is now capable of learning from its own memory without input from its programmers, the team responsible has revealed. The DeepMind researchers have developed a Differential Neural Computer (DNC), a new hybrid system that connects the AI’s neural network to a large data storage bank. The AI can now access the information from this data storage and learn from it independently. Effectively, DeepMind can now both read and write data at will.

The article on DNCs is #DeepMind's third @Nature paper in just over 18 months! Pretty unprecedented in computer science! #OptimizingScience

— Demis Hassabis (@demishassabis) October 12, 2016

The researchers tested DeepMind’s capacity for independent learning by tasking it with navigating the London Underground subway train network.

According to DeepMind’s blog:

“We wanted to test DNCs on problems that involved constructing data structures and using those data structures to answer questions. Graph data structures are very important for representing data items that can be arbitrarily connected to form paths and cycles. In the paper, we showed that a DNC can learn on its own to write down a description of an arbitrary graph and answer questions about it. When we described the stations and lines of the London Underground, we could ask a DNC to answer questions like, “Starting at Bond street, and taking the Central line in a direction one stop, the Circle line in a direction for four stops, and the Jubilee line in a direction for two stops, at what stop do you wind up?” Or, the DNC could plan routes given questions like “How do you get from Moorgate to Piccadilly Circus?””

The DNC development points to a future in which AIs can generate potential answers to questions without having every possible outcome already input into its memory, merely by deduction based on previous experience and processing basic data.