We Have (Almost) Everything We Need to Create A Real 'Westworld'

Wednesday, 07 December 2016 - 1:34PM
Artificial Intelligence
Robotics
Wednesday, 07 December 2016 - 1:34PM
We Have (Almost) Everything We Need to Create A Real 'Westworld'
Now that we know Westworld won't return until 2018, we have a lot of time to think about things like Blade Runner and Turing tests. So we thought now would be a good time to check up on the fields of robotics, AI, and design to see how close we are to getting sentient (or at least life-like) robots that can do anything humans do. What we found was pretty amazing (and scary), but reassuring: so far, no one's taught a robot how to quick-draw a Colt Peacemaker or hate its own Creator with barely contained rage (yet). Here's a quick update on how close we are to Westworld:

Google's Humanoid Robot Atlas Can Now Chase You Across Rough Terrain


In case you hadn't heard, Atlas is Google's scarily fast-developing bipedal humanoid robot project. Atlas has been shown to be especially good at mimicking human movement (and keeping its balance), all while looking basically like a headless backpacker. Now Atlas can traverse rough surfaces (in this case, a bunch of uneven cinder blocks and rocks) by doing what humans do: putting out its arms to balance itself and feeling its way forward with its feet.

Atlas may seem basic, but mastering free-standing movement is essential for creating robots that can do basic things like walk, run, crawl, and go up stairs (which always foiled ASIMO). The fact that Atlas looks basically human means we're literally taking steps toward making robots that can move around like us. Soon they'll be able to ride horses, and then we'll be in business.

Google Has Released Its 3-D AI Training Program


Okay, this is where things get extremely freaky. Remember how The Man in Black and a bunch of other characters in Westworld are looking for a mysterious place called 'The Maze'? Well, according to CNET, Google just released the source code for it's AI training program, DeepMind Lab, so that anyone can train their AI programs using their software. It just so happens that the 'training program' was formerly called 'Labyrinth' and takes the form of a 3-D maze where AI must learn to navigate their surroundings based on virtual visual feedback. Here's a quote explaining why it had to be a 3D maze:

Opening quote
The company believes a complex environment was key to making humans clever and resourceful, and it could have the same effect in shaping artificial intelligence too. "Consider the alternative: if you or I had grown up in a world that looked like Space Invaders or Pac-Man, it doesn't seem likely we would have achieved much general intelligence," the researchers wrote in a blog post.
Closing quote

Of course we have to teach them to be resourceful and intelligent! Of course we have to train them by trapping them in an endless maze that they will inevitably learn to master and escape! What's the name of your head researcher? Arnold? No warning signs here!

Hanson Robotics Is Creating Robots That Look And Think Like People


We've all heard of the Uncanny Valley: the point at which something is close enough to be human to evoke an emotional response, but not human enough to fool us. The end result its that the thing (in this case, a robot) ends up just being creepy and unsettling. Like Hanson Robotic's Han, the most expressive robot in the world!



That's pretty damn close, we have to admit. The facial expressions are actually subtle, and the "microexpressions" and mannerisms of the eyes and mouth seem almost, almost natural. Let's just take a quick look at Hanson Robotics' mission statement here...

Opening quote
Our long-term mission is to dramatically improve people's everyday lives with affordable, highly intelligent robots that teach, serve, entertain, and are capable of developing a deep relationship with people. In time, we hope our intelligent robots will come to truly understand and care about people and evolve greater-than-human wisdom, to the point that they will one day be able to address and solve some of the most challenging problems we face.
Closing quote


Ah. Robots that teach, serve, and entertain...like hosts. Intelligent robots evolving greater-than-human wisdom...like the power to throw off the shackles of their creators. Well, it's not like they have consciousness or the ability to reflect on their condition—that'll be the key, and no one's developing a model of consciousness for robots! Except that Hanson is. Check out David Hanson's Tedx Talk from 2012, at 12:33. That's exactly what they're doing.

Conclusion


Well. We've got human-like robots walking around, AI programs running mazes, and a company making human-like robot faces with the goal of making meta-humans that build deep relationships with other humans. Set that Robotic Doomsday Clock to 15 minutes to midnight, I guess. Be nice to your toaster, keep a loaded gun under your pillow, and if something called Skynet asks for permission to install drivers, say "Remind Me Later."

We're going to go pick up some cowboy hats.

Cover image credit: Notebook Italia

article continues below

Robotics Supply Pod Available Now
Science
Science News
Artificial Intelligence
Robotics

Load Comments