Westworld looks set to be one of the most exciting science fiction series television has ever seen, but could we ever experience it for real in our lifetime?
Imagine a world where you could act out the dreams you’ve been nurturing ever since you watched The Good, The Bad and The Ugly as a young child. You could walk through the swinging doors of a saloon, drink some whisky, and then whip out your revolver and fill a bandit with lead. Sound good? Well, that’s the world offered in the new science fiction series set in Westworld: a futuristic theme park where humans can pay to live out their gunslinger fantasies in an artificial Wild West inhabited entirely by humanoid machines who exist purely to play the supporting cast.
Of course in the show there are some rather unsavoury visitors to the theme park, and some just as sinister malfunctioning machines who turn against them, which looks like the beginnings of great entertainment. But the idea still got us excited; an immersive theme park filled with robots! That’s something I’m sure we’d all enjoy. So just how far away are we from achieving place for real? It wouldn’t be the first time science “fiction” from past films has become a reality later on (although Marty McFly’s hovering skateboard has still somehow managed to elude us) so it may not be beyond belief that we could build a real life Westworld one day.
What makes Westworld feel so real (and ultimately so dangerous for some in the show) is that its robot inhabitants walk, talk and act just like humans as they make their scripted way around the constructed artificial world. Building the world would certainly be the easy part, even though it’d take up a fair bit of space. Canada’s Devon Island is completely deserted and formed of mostly barren terrain so could be a great building spot, but as it’s near the North Pole it wouldn’t be the easiest place to visit for the casual tourist. Maybe somewhere near Disney world instead then.
As for the humanoid bodies, our modern research direction is suggesting that it may be easier to make synthetic humans, rather than constructing the robotic frames coated in human-like skin that they use in the show. Recent advancements in regenerative medicine could see this as a reality sooner than we might think. Harvard scientists have seen significant progress in their aim to grow entire human organs – achieved by attaching stem cells to a matrix scaffold taken from another mammal (such as a rat or pig) before growing the organ in a bioreactor – and the principle is being developed for multiple different organs. This research is just one example of many that suggests growing or augmenting human tissue in the lab could lead to synthetically created life faster than we’ll be able to build humanoid robots. But to be considered machines the future Westworld inhabitants must have machinery, and that could come in the form of their intelligence. And artificial intelligence may be arriving on Earth even sooner than the synthetic bodies – an eventuality many science fiction fans both crave and fear.
So just how soon will we be building intelligent machines? We consulted with Oxford University computer science professor Nando de Freitas for his say on AI
Intelligence is a tough concept to pin down. Is someone, or something, intelligent because they can multiply 29 x 18 in their head? What about emotional intelligence – working out if someone is sad, angry or faking an emotion? There’s a fair few people who could solve the above sum (it’s 522, if you were wondering) but wouldn’t have a clue what to do when someone starts bawling their eyes out in front of them. All this makes the field of artificial intelligence a thorny path to tread.
British mathematician Alan Turing, most famous for his codebreaking work at Bletchley Park during World War II, got people thinking about computers as a tool for thought, rather than mere calculations. But where are we now in terms of creating computers and robots that can think, talk and perform tasks like humans? After recent advances we are close to a breakthrough in AI, but have a little way to go yet before engineering our artificial heroes and villains of the Western-styled theme park.
Oxford University’s professor of Computer Science, Nando de Freitas, has spent a lot of time studying the brain, trying to work out what goes on without us knowing it: “There’s an area in the brain called the hippocampus, which is fascinating. In a rat, particular neurons will fire when it is travelling in a certain direction, but only for that direction. However, if it is in a different part of the room, a different neuron fires. That is how the rat knows where it is in the room. Each neuron is connected to the visual cortex, where we store images, and the auditory cortex, where we store sequences. Each neuron represents a location in the world and fires when you are there. Every time you excite a neuron in the hippocampus, it fires a certain set of neurons representing an image in the world, meaning, for example, you can now imagine your way home. Right now, we are trying to work out how to do that. We take intelligence for granted. You aren’t aware of a lot of what’s going on in your brain, which is why it’s so hard to reverse engineer it.”
The argument that a robot would be unable to be truly intelligent until it can feel emotions like a human is easy to refute. “Emotions are one of the easiest things to reproduce,” argues De Freitas. “You don’t need to build something as intelligent as a human to get an emotional response. If you were to poke the amygdala with a needle, you’ll get an emotional response. That’s because the amygdala is part of the old brain, which we share with rats, mice, cows, pigs and lots of other animals. The new brain – areas like the neocortex – is where we do our higher level of thinking.”
So if scientists aren’t looking at developing robots that get sniffly at The Notebook, how are they attempting to create the next generation of thinking robots? After all, we have had Deep Blue, then Watson, then AlphaGo. What hurdles are they yet to overcome?
“I see intelligence as being able to interact with an environment and do the right thing,” continues De Freitas. “Humans are able to plan their actions and engage in counterfactual reasoning, which is a fancy way of saying ‘what if’ reasoning. That is being able to perform an action and ask yourself: ‘What would happen if I did this other thing?’ Robots are much smarter than us. They can perform logic and mathematical tasks much quicker. However, we can go from observing sequences in the world and build representations of them in our brains. This is what we are now trying to achieve with AI.”
De Freitas is heavily involved with the development of Deep Learning, a programme which looks to replicate the human brain’s ability to not only see an image but understand it as well – something major technology corporations like Google, Amazon and Facebook are keen to exploit.
“Deep Learning tries to get robots to build representations of the world and operate on those representations to build sequences in their mind. Then we want them to learn to construct different sequences. It’s like if you take all the videos on YouTube, cut each video into ten-frame chunks and cut and paste them into new movies. The next step is to imagine alternative scenarios to what is put in front of them. We want computers to learn abstract representation about their environment and then think about their environment and the cause and effect of their actions.
“All the big search engines already use this tech. For Facebook, that means learning about users from all the data they input. You can learn a lot from the data that exists out there – even their IQ. They could use this data to start recruiting, or even become a life coach. No psychologist has ever had access to this amount of data. I talked to Mark Zuckerberg about this a few months ago. There’s a reason why he’s investing in this.”
Google is also taking a close interest in the possibilities AI brings to the table. The company’s reported £242 million ($400 million) acquisition of DeepMind, a London-based AI company, has already started to bear fruit. This year their AI AlphaGo managed to defeat a world champion of Go – one of the world’s most complex boardgames – using a sophisticated neural network that some experts didn’t believe would be possible to create for another 10 years.
This coming decade has been predicted as a pivotal point for the development of AI. Many of the top technology companies around the globe have recently announced that they will be collaborating on how to best regulate and develop the AI of the future. So AI systems may really be here soon and find themselves operating devices and working inside companies around the world. And who knows, maybe they’ll be found in a Western-styled theme park too.
The first episode of Westworld will be on at 9pm, 4th October on Sky Atlantic in the UK.
Written by the How It Works team.
Discover more amazing features in the latest issue of How It Works. It’s available from all good retailers, or you can order it online from the ImagineShop. If you have a tablet or smartphone, you can also download the digital version onto your iOS or Android device. To make sure you never miss an issue of How It Works, make sure you subscribe today!
You might also like: