We all love a good story. And with so many unknowns, it’s easy to see why artificial intelligence is at the center of so many tall tales. In Hollywood, movies like The Terminator, The Matrix, or The Stepford Wives all follow a common trope: AI at odds with humanity, with AI often getting the short end of the stick.
With AI, it’s important to separate fact from fiction. What the technology actually is and what it can currently do is wildly different from how it looks on the big screen. So, to help you get a better understanding of this science, I’ll be kicking off a new blog series about everything the movies get wrong about AI, starting with the movie Transcendence.
Released in 2014, “Transcendence” tells the story of a renowned scientist and AI authority, Dr Will Caster. Due to his work in developing a “sentient computer,” Caster is targeted by an anti-technology extremist group called RIFT (Revolutionary Independence From Technology). During a series of coordinated attacks and assassinations on artificial intelligence labs across the United States, Will is shot and it’s later discovered that the bullet was laced with polonium, a “rare and highly radioactive metal,” giving Will just a few weeks to live.
Soon, however, it is learned that one of their recently assassinated colleagues figured out how to “upload” the consciousness of a rhesus monkey onto a computer. With the help of his computer scientist wife Evelyn and their reluctant best friend and fellow scientist Max, the same is done to Will’s consciousness, allowing him to live on after the radiation kills his body. Now harnessing the power of artificial intelligence, Will is quickly deemed to be a threat to humanity, and as such, the FBI, particularly when he begins to use his newfound abilities in nanotechnology to seemingly control human beings.
So, what does Transcendence get wrong?
1. Basic AI is more advanced that humanity’s most brilliant minds.
At the beginning of the film, we see Will giving a talk on “singularity” or as he likes to call it “transcendence.” There, he states that the combined intelligence of all the neuroscientists, mathematicians, engineers, and hackers in the auditorium “pales in comparison to even the most basic AI.”
Now listen, as a writer, I understand the goal here. AI is the enemy, so Will telling his audience that even experts with years of higher education and experience in some of the most complex fields in the world can’t compare to a computer is meant to set us up for “AI Will’s” later invincibility. In real life though, as I noted here and here when we speak about the “most basic AI” today, we’re referring to narrow AI, which focuses on one task. At its best, these algorithms can do their single tasks very well, allowing it to augment humans. Computer vision, for instance, has been used to develop solutions that moderate visual content, identify faces and teach us new languages. That said, in all of these use-cases, computer vision is still really doing one thing: allowing machines to “see.” Humans, on the other hand, like those Will says can’t compare to basic AI, see and remember new faces, hold coherent conversations, and solve scientific problems every day, often without any conscious effort. Scientists have yet to develop an AI that can do all of that.
So, while in Transcendence, Will’s AI is compared to a god and seems to be as unconquerable as Zeus himself, in actuality, AI is still well behind the mortals who worshipped and feared the ancient Greek deity.
2. General AI is around the corner.
In this movie, set in the near future (or present since it’s 4 years old), general AI seems to have been actualized. Will, for example, has already developed PINN (short for physically independent neural network). PINN is a supercomputer that we really only see do two things well (facial recognition and natural language processing), but I digress. In the film, it is supposed to be relatively conscious, as shown when Will’s mentor asks if it can prove it is “self-aware” (i.e. if it’s general AI), answering with some Siri-esque snark “That’s a difficult question. Can you?”
AI pioneer Alan Turing developed his test for determining whether an AI is truly intelligent in 1950. Almost 70 years on, it’s still not passed the test, and no one can give a definitive answer as to when it will. Building an AI solution to tackle even the world’s most fundamental problems is not as simple as Will and company make it look. It’s why companies have to think long and hard when deciding whether to build or buy and even if they settle on the latter, who to buy from. That’s not to say there hasn’t been any success with AI. Many industries have seen the benefits of some AI in action. However, when we talk about these successes, we need to acknowledge they have come after decades of research and much trial and error.
In the movie, they can just use a few electrodes to replicate the consciousness of monkeys and humans.
“He [referring to one deceased colleague] was decoding synapses in real time,” Will says they discuss “uploading” Will’s brain to PINN. “But with the right input algorithm...” And that’s the thing. While Evelyn insists that “the brain is just a pattern of electrical signals that they can upload to PINN” “like a song or a movie,” part of why developing general AI has been so challenging is the complexity of the human brain. Experts only recently created the right algorithms to teach computers how to identify dogs versus cats as well as humans can. It’s likely going to be some time before they actually decode how information floats from neuron to neuron, let alone develop algorithms that can replicate this.
3. The promise and peril of AI
The main antagonist of the movie, RIFT, is run by a former computer scientist turned anti-technology vigilante named Bree. She tries to entice Max into joining RIFT by playing on his concerns about the ethics of developing a sentient computer, saying “most men of science are blind to it.”
Science is a broad field divided into many independent branches, but let’s assume Bree is referring to men of computer science. AI is just one of several subfields under this umbrella, so many of these “men of science” are likely not well-versed when it comes to AI (which itself has different subspecialties). The lack of understanding from these men and the general public is expected, as AI is still very much in its infancy. Their concerns are also understandable but likely influenced by movies like Transcendence that tell us that we’re steps away from creating a sentient computer that’s going to go full Agent Smith, hellbent on destroying the human race.
Within the field of AI, there is some debate over its promise as a technology and concerns over how it can be abused. That said, much of the subject is still so unknown to so many of us, that a lot of the concern is the result of the boundless human imagination that made AI possible in the first place. As AI continues to advance globally, we will need to have productive discussions around it if we are to innovate responsibly and develop informed regulations for it. So, while AI can be entertaining in film, it is crucial for companies like ours to prepare the general population for this technology by providing them with facts to go alongside these fictional portrayals.
4. AI will control, then destroy, well, everything.
After Will is successfully turned into “AI Will,” he becomes a power-hungry AI. Against Max’s, er, will, Evelyn connects “AI Will” to the internet and soon “AI Will” knows, and controls, everything. God-style. It’s a playback to the beginning of the movie when Will’s would-be assassin accuses him of trying to create a god with his sentient computer, and Will, smugly if you ask me, smiles and responds, “Um, isn't that what man has always tried to do?”
The debate on the existence of god/s is only relevant here because “AI Will” becomes Zeus within the first few days of his existence. A few moves on the net and Evelyn is a multimillionaire, who at Will’s instruction, buys a town in the middle of the desert and turns it into the grounds of their research facility. There, “AI Will” uses his newfound expertise in nanotechnology (something that is said to be “decades away” from fruition earlier in the movie, but “AI is god,” remember?) to heal life-threatening injuries, paraplegia, and blindness in seconds. Unfortunately, he also uses these “nanoparticles” to turn all who come to him into invincible cyborgs, subject to his control. Oops, I suppose, happens to the best of us.
Undoubtedly, AI can augment humans and, in fact, already has. Can it turn us into Herculean androids subject to the control of some general artificial intelligence that’s gone a little mad with power? I actually have no idea. What I do know is that “AI Will” (seemingly) using his powers to control humanity is well within the compounds of this trope. In full Zeus mode, “AI Will” is controlling the weather and regenerating anything that is destroyed or harmed with nanotechnology: limbs, solar panels, even himself. We have yet to develop an AI that can adequately write a good book, so might be a while before it’s controlling the universe.
So, in conclusion, while Transcendence was no blockbuster, it does play on a few common stereotypes about AI in fiction. General AI may be coming, but we are still very much in the “narrow AI” phase of its development, and this is to our benefit. As our CEO noted here, AI is still very much under human control. What it becomes is up to the programmers in the field and how humanity uses the technology.