ZME Science
No Result
View All Result
ZME Science
No Result
View All Result
ZME Science

Home → Research → Technology

Google’s Artificial Intelligence algorithm just navigated London’s underground without any prior knowledge

Google's DeepMind algorithm just got one step closer to behaving like a human.

Mihai AndreibyMihai Andrei
October 17, 2016
in News, Robotics, Technology
A A
Share on FacebookShare on TwitterSubmit to Reddit

Google’s DeepMind algorithm just got one step closer to behaving like a human.

Human smart

DeepMind has gotten scarily smart in recent months – and we’re not just talking “computer smart,” but more like “human smart.” Just in 2016 the algorithm defeated the world champion at Go – something which was thought to be impossible as Go is hugely more complex than chess – by not only learning from previous games, but also by developing its own, completely novel strategies.

You see, unlike IBM’s Deep Blue or Watson, which were developed with a predefined purpose, DeepMind has no purpose. It simply learns from its surroundings, takes in information, and develops its own approach to different tasks. Technically, that’s called deep learning on a convolutional neural network. For instance, without altering the code in any way, DeepMind quickly learned how to play simple arcade games such as Space Invaders just by observing the game.

But that’s not all the algorithm does. Google let it in charge of its electricity management, and saved a whopping 15% of all costs – they believe many other companies and even households could benefit from it. Currently, the algorithm is involved in two other projects: with the Moorfields Eye Hospital, to search for early signs of diseases leading to blindness, and with the University College London Hospital to automatically differentiate between healthy and cancerous cells in the head and neck areas.

DeepMind is so smart that engineers working with it are considering installing a kill switch in case the algorithm goes rogue. But in the meantime, they’re teaching it how to ride the subway.

The London tube

The London subway, or the Tube as locals call it, is one of the most complex in the world. As any tourist will gladly tell you, it’s not easy to navigate it and you can easily get lost in the traffic and the rush. DeepMind didn’t know anything about the subway or about navigating public transportation. To make it navigate a subway model they had to think outside the box and so they did.

They developed a neural network which works with an external memory. The combination allows the neural network not only to learn, but to use memory to store and recall facts to make inferences like a conventional algorithm. This is exactly the type of technology which allows it to learn how to accomplish tasks without previous knowledge. Again – for a pre-programmed algorithm, this would be an easy task. But for an algorithm without any related programming, it’s a whole new ball game.

RelatedPosts

Another first: NASA’s Perseverance rover extracts oxygen on Mars
Dust devil on Mars, as seen by Opportunity Rover
Notre Dame’s 3D virtual reconstruction offers hope for the cathedral’s full restoration
Tesco, world’s second largest food retailer, to give all unsold food to charity

This has intriguing consequences. Sure, going from Picadilly to Westminster is not going to change the world, but it does indicate that algorithms are starting to become more and more human-like. Where DeepMind will take off next is anyone’s guess.

The paper was published in Nature on 12 October. Journal Reference: Hybrid computing using a neural network with dynamic external memory.

ShareTweetShare
Mihai Andrei

Mihai Andrei

Dr. Andrei Mihai is a geophysicist and founder of ZME Science. He has a Ph.D. in geophysics and archaeology and has completed courses from prestigious universities (with programs ranging from climate and astronomy to chemistry and geology). He is passionate about making research more accessible to everyone and communicating news and features to a broad audience.

Related Posts

Archaeology

A 2,300-Year-Old Helmet from the Punic Wars Pulled From the Sea Tells the Story of the Battle That Made Rome an Empire

byTibi Puiu
1 hour ago
Health

Scientists Hacked the Glue Gun Design to Print Bone Scaffolds Directly into Broken Legs (And It Works)

byTibi Puiu
2 hours ago
Science

New Type of EV Battery Could Recharge Cars in 15 Minutes

byMihai Andrei
2 hours ago
Microbiology

How Much Does a Single Cell Weigh? The Brilliant Physics Trick of Weighing Something Less Than a Trillionth of a Gram

byTibi Puiu
3 hours ago

Recent news

A 2,300-Year-Old Helmet from the Punic Wars Pulled From the Sea Tells the Story of the Battle That Made Rome an Empire

September 15, 2025

Scientists Hacked the Glue Gun Design to Print Bone Scaffolds Directly into Broken Legs (And It Works)

September 15, 2025

New Type of EV Battery Could Recharge Cars in 15 Minutes

September 15, 2025
  • About
  • Advertise
  • Editorial Policy
  • Privacy Policy and Terms of Use
  • How we review products
  • Contact

© 2007-2025 ZME Science - Not exactly rocket science. All Rights Reserved.

No Result
View All Result
  • Science News
  • Environment
  • Health
  • Space
  • Future
  • Features
    • Natural Sciences
    • Physics
      • Matter and Energy
      • Quantum Mechanics
      • Thermodynamics
    • Chemistry
      • Periodic Table
      • Applied Chemistry
      • Materials
      • Physical Chemistry
    • Biology
      • Anatomy
      • Biochemistry
      • Ecology
      • Genetics
      • Microbiology
      • Plants and Fungi
    • Geology and Paleontology
      • Planet Earth
      • Earth Dynamics
      • Rocks and Minerals
      • Volcanoes
      • Dinosaurs
      • Fossils
    • Animals
      • Mammals
      • Birds
      • Fish
      • Amphibians
      • Reptiles
      • Invertebrates
      • Pets
      • Conservation
      • Animal facts
    • Climate and Weather
      • Climate change
      • Weather and atmosphere
    • Health
      • Drugs
      • Diseases and Conditions
      • Human Body
      • Mind and Brain
      • Food and Nutrition
      • Wellness
    • History and Humanities
      • Anthropology
      • Archaeology
      • History
      • Economics
      • People
      • Sociology
    • Space & Astronomy
      • The Solar System
      • Sun
      • The Moon
      • Planets
      • Asteroids, meteors & comets
      • Astronomy
      • Astrophysics
      • Cosmology
      • Exoplanets & Alien Life
      • Spaceflight and Exploration
    • Technology
      • Computer Science & IT
      • Engineering
      • Inventions
      • Sustainability
      • Renewable Energy
      • Green Living
    • Culture
    • Resources
  • Videos
  • Reviews
  • About Us
    • About
    • The Team
    • Advertise
    • Contribute
    • Editorial policy
    • Privacy Policy
    • Contact

© 2007-2025 ZME Science - Not exactly rocket science. All Rights Reserved.