Check out all the on-demand sessions from the Intelligent Security Summit here.
It seems like every television season plays host to a single show that takes over our cultural consciousness by coating modern fears in fantastical drama.
Game of Thrones taps into the cutthroat nature of modern politics, while The Walking Dead plays into our ever-present fear of a worldwide contagion. This year, with Elon Musk’s dire warnings about the dangers of artificial intelligence, Siri and Alexa permeating our personal lives, and self-driving cars threatening to run down groups of children, Westworld is posed to be that show.
Westworld is complex. It’s that complexity, interwoven with mysteries and plot twists, that produces the hours of online speculation and wild fan theories that are the hallmark of anything worth watching.
Yet, because Westworld‘s brand of complexity is intrinsically rooted in technology, some of the more technically inclined fans have a sizable leg up on other fans when it comes to understanding this futuristic fairy tale. So we’re going to break it down for the non-techie cohort. What are the inventions of the show, how do they work, and how close are we to achieving them today?
Intelligent Security Summit On-Demand
Learn the critical role of AI & ML in cybersecurity and industry specific case studies. Watch on-demand sessions today.
The premise of Westworld is deceptively simple: a Wild West theme park allows Guests to take on roles and engage in storylines with Hosts, or androids (a word for a robot that seems human). The Hosts can improvise based on what the Guests do, and over the course of the show they begin to achieve self-awareness. It may seem impossible that such a thing could happen in 2016. And while that’s true, some of these innovations are closer than you might think.
1. 3D printing
In the gorgeous opening credits of the show, with musical stylings provided by none other than Games of Thrones composer Ramin Djawadi, we see the process of one of the android bodies being created. A three-dimensional humanoid shape emerges from a strange white bath. We see the iris of an eye being carefully, laboriously printed one strand at a time by a robotic arm not unlike the ones you find toiling away inside a modern automotive assembly line; bones, muscles and tendons are slowly layered into being. We all know that 3D printing is possible, but could we actually create a human body this way?
The inspiration for the 3D printing technique used in Westworld comes from a well-established technology called stereolithography (SLA) 3D printing. According to Wikipedia, SLA is an additive manufacturing process that works by focusing an ultraviolet (UV) laser onto a vat of photopolymer resin (the white liquid seen dripping from the androids during assembly). For this to work in the way the show wants it to, both the speed and accuracy of the method would have to increase dramatically. Without the addition of nanotechnology (the ability to assemble structures atom by atom), it’s doubtful we could create complex enough structures to mimic the human body. Furthermore, rather than a single ubiquitous material, one assumes that an electrically conductive substance would need to be embedded (the “nerves”) in order to actually signal the synthetic muscles (don’t worry, they’re working on that).
There are also people printing 3D organs, but there’s no indication (yet) that the Hosts actually have organs. They have blood (as evidenced by all of the times they’ve been shot, plus that horrifying head smashing), but chances are they’re made of a flesh substitute. A more realistic powersource would be a small, ultra energy dense battery, similar to what a company called Ampirus is working on now.
Our Take: Plausible, mid-far future
2. Robots that seem human
The Hosts move so humanly that the opening scenes of the show rely on tricking the audience into believing that Teddy is one of the human Guests; the reveal partway through the episode that he is in fact one of the androids resounds especially with fans of the original movie.
This is one of the areas from which we are currently the most removed. While we can create androids that look incredibly real, they are given away instantly by their movements. Judging by the latest robotic competitions and what has come out of Boston Dynamics, the machinery needed to actually walk is still very far from what we seen on the show. However, the current bottleneck is computational power, which grows exponentially each year.
Our Take: Plausible, mid future
3. Intelligent machines
As a slight departure from the original work, show creator Jonathan Nolan draws our attention to the nuances of human-AI interaction rather than mechanical malfunctions. If you’ve talked to any chatbots on Facebook Messenger lately, or had a conversation with Tay on Twitter, you know that we’re still a ways off from any kind of true general intelligence. But the thing about the androids of Westworld is that, at least at the beginning, they’re not truly intelligent — they’re just programmed so thoroughly that upon a first meeting, the average person can’t tell the difference.
However, Hosts are still far more advanced than a simple scripted chat interface — they’re capable of improvisation, and the way they’re left to “practice” with each other suggests a form of machine learning. In fact, the need to “reset” the Hosts, erasing past memories, implies they learn very well, and could otherwise escape their pre-programmed narrative loops. The Shakespearean phrase we see in Episode 2 seems to be a trigger that gives the Hosts the ability to access their prior databanks, which are seemingly “erased,” but actually still present. This is also how a hard drive on your computer works — when you delete data it isn’t actually removed, it’s simply treated as empty space to the operating system, and only erased if new data overwrites it.
By current standards, we are still a long way from a true general AI, but we can break down the Hosts’ abilities into a set of practical (and potentially solvable) problems. We know from the show the Hosts respond to voice activation commands as well as physical interfaces, such as tablets and computers. So at minimum we need: advanced speech, object, and facial recognition and comprehension; a semblance of world knowledge; and an improvisation and reasoning engine tying it together.
You may be surprised to know that we have already achieved better-than-human speech recognition. Interestingly, because humans only achieve 80 to 85 percent accuracy, improving this technology would actually make them seem less real!
Our Take: Already here
4. Image recognition
Anyone with an internet connection will know that Google, Facebook, and Apple have made huge advancements in image recognition — just look at Google Photos’ ability to automatically group your photos by person, even as they age. Hosts need to detect objects as well, and interestingly advancements in this area might come from self-driving cars. It turns out that Tesla’s AI is able to leverage the fact that the car is moving by looking at the way the object changes given that movement being mobile; the Hosts will have the same advantage.
Our Take: Definite, nearly here
5. Infinite knowledge
World knowledge is required in order to have real comprehension. It isn’t enough to be able to say: This is a picture of an apple, and this is a picture of an orange. You need to know that we eat oranges for breakfast, that apples are symbolic of original sin, and that no one puts either into a garden salad.
Researchers at IBM have made this kind of knowledge their mission, and IBM Watson is the result of years of research on the subject. You’re probably familiar with Watson from when it won a game of Jeopardy against human players. Currently that database of knowledge needs to be programmed, but with enough time, we’ll start to see androids that can access and understand a nearly infinite pool of knowledge.
Our Take: Definite, nearly here
There is a wide rift between accessing programmed responses and being able to improvise. Programmed responses are simple — we can create an android today that could put on a play with another android, using pre-programmed scripts. What is much harder is giving them the ability to improvise.
Current chatbots give the illusion of improvising by selecting answers based on a list of things you might say, a “choose your own adventure”-style story. However, we have given computers the ability to improvise, to a certain degree. When AlphaGo (Google’s deep learning experiment) beat a human master of the game Go, it would have been impossible to program it with every possible move (there are more potential moves in Go than stars in the universe). Instead, researchers used a deep neural net approach to let AlphaGo teach itself how to play, by studying hundreds of thousands of other games.
The downside to this approach is that researchers have no knowledge or control of how the machine is making those improvisational leaps. Sounds kind of like Westworld, doesn’t it?
So if we’ve achieved speech parity, and image recognition is on point, and neural nets allow us to teach machines to improvise … why don’t androids exist today?
The hangup is that the computing power (and data) needed to create such a network is immense. We have to assume that in Westworld‘s future, they’ve had a massive breakthrough, perhaps via quantum computing or a massive worldwide collaborative computing effort (similar to Worldwide LHC Computing Grid, which is expected to compute 50 petabytes of data). So our take on this one assumes that sometime, computing power is no longer the bottleneck it is today.
Our Take: Plausible, near-mid future
While the show focus on the androids and adventures of the Westworld theme park, the world extends far beyond those borders, and technology has continued to advance in all areas of life. Here are some minor technologies we glimpse throughout the show.
In a world based in technology, one of the first questions we have to ask is: What do computers look like? In a short time computers have evolved from giant machines that span entire rooms to tiny supercomputers we carry around in our pockets (smartphones).
In Westworld, we see computers both large and small, indicating that they’re still constrained by size the way we are today — we can make incredibly minute computers, but the larger they are, the better their computing power. So, big computer banks span Westworld’s Control Room, while smaller tablets are used to interface with the androids during maintenance sessions.
The coolest part of the tablets is that they fold up to be placed in your pocket. Technology like that might seem farfetched, but in actual fact, Lenovo has just announced a foldable phone that will double as a tablet. Its scheduled release date? 2017.
Our Take: Already here, essentially
The park takes up a massive amount of real-world land, and part of providing countless new entertainments for Guests is constant change. That means not only new Hosts and new storylines to participate in, but entirely new lands to explore.
In the fourth episode we see an Ultra-Digger churning up the land, breaking down a restaurant that’s been there for forty years to make way for a new storyline and landscape. Fortunately or unfortunately, this technology is very real, and is in use today in coal mines around the world.
Our Take: Already here
The mystery continues
Westworld toys with a recurring theme: “Imagining something is there when it is, in fact, not,” as Bernard scoffingly tells Elsie after an android carves what looks like Orion’s belt into a series of statues, humans are “pattern-seeking, story-telling animals.”
We personify everything around us. Throughout history our species has imagined sentience in animals, plants, even weather patterns. Intellectuals, especially scientists, are trained to fight that instinct, and it seems to be that very battle that makes Guests and creators alike blind to the emerging intelligence of the Hosts they have created.
Only time will tell if the androids of Westworld are truly intelligent or just reflecting back at us. But as we have seen, the future Westworld presents is all too possible. The lesson to be taken from this is clear: We should prepare for this future’s swift and inevitable arrival. We can start by thinking hard about the nature of intelligence, artificial or otherwise, and perhaps even put ourselves in the Hosts’ shoes.
VentureBeat's mission is to be a digital town square for technical decision-makers to gain knowledge about transformative enterprise technology and transact. Discover our Briefings.