Stories about artificial intelligence have been with us for decades, even centuries. In some, the robots serve humanity as cheerful helpers or soulful lovers. In others, the machines eclipse their human makers and try to wipe us out.
“The Creator,” a sci-fi film that hits theaters Friday, turns that narrative around: The United States is intent on wiping out a society of androids in Asia, afraid the artificially intelligent beings threaten human survival.
[The new Bing told our reporter it ‘can feel or think things’]
Do any of these stories reflect our real-life future? How have they influenced the sometimes awe-inspiring technologies being developed today?
The Washington Post compiled a list of archetypes, characters and films that have been most influential in shaping our hopes and fears about artificial intelligence. We spoke to computer scientists, historians and science fiction writers to guide our understanding of how AI might evolve — and change our lives.
The Killer AI
Meet your worst nightmare: The AI system that achieves sentience and seeks to destroy — or enslave — humanity.
In 1984’s “The Terminator” and its 1991 sequel, the villain is Skynet — a highly advanced computer network created as America’s first fully automated defense system, with control over all battle units. Powered up on Aug. 4, 1997, Skynet becomes self-aware 25 days later and starts a nuclear war that annihilates billions of people. It then builds an army of robots to kill the survivors.
This is the AI apocalypse that haunts the dreams of some scientists, who are racing to create “artificial general intelligence” — an AI system that’s as smart as a human — in hopes of shaping the technology to share our morals and serve humanity. Others call this a fool’s errand: “We don’t know how to train systems to be fully sensitive to human values,” said Mark Riedl, associate director of the Georgia Tech Machine Learning Center.
Still, others reject the idea of killer AI entirely, saying fears of a real-life Skynet are overblown.
Even a supercomputer like HAL 9000 from “2001: A Space Odyssey” (1968) is a very long way off, experts say. In the Stanley Kubrick classic, HAL is able to “reproduce” or “mimic” many activities of the human brain “with incalculably greater speed and reliability” as it controls operations on a space mission to Jupiter. HAL begins killing human crew members when they discuss turning it off — the same threat that motivates Skynet.
‘I’m sorry, Dave’
Stephen Mihm, a professor of history at the University of Georgia, called rebel AI “a perennial anxiety, and a distraction,” that “probably speaks more to the anxiety about the arrival of a new technology than it necessarily says about the likelihood that the technology will turn on us.”
Policymakers are not waiting to find out. Congress is working to develop guardrails for AI development, considering a new agency to regulate the industry and weighing legislation that would prohibit giving an AI system final say over U.S. nuclear strategy.
Other homicidal AI characters:
The AI Lover
Swipe left on this romantic prospect: The AI love interest rarely works out.
In “Her” (2013), a lonely writer named Theodore (Joaquin Phoenix), falls for Samantha, a disembodied but hyperintelligent computer assistant voiced by Scarlett Johansson. Theodore is entranced by Samantha’s enthusiasm to grow and learn, and they develop a romance. He is heartbroken when Samantha evolves to communicate with other AIs, who leave him and the rest of humanity behind.
‘Are these feelings even real?’
Over the past year, people have become strongly attached to — and even married — AI chatbots powered by large language models, the technology behind ChatGPT. Scholars say this is possible because of the “Eliza effect” — our erroneous perception that some AI systems possess a deep inner life.
“It doesn’t take much for people to fall in love with anything,” said Isabella Hermann, a German political scientist who studies the intersection of AI and science fiction. “We anthropomorphize things and talk to things — and it doesn’t really need to be that intelligent.”
In the movies, it’s never quite clear whether AI romantic partners truly “feel” or are just programmed to make humans think they do. Some pine to be fully human, like the android Data (Brent Spiner), who attempts multiple romantic relationships throughout the television series “Star Trek: The Next Generation” (1987-1994).
While there’s no evidence that the current romantic chatbots are self-aware, critics say they can be addictive. A Google search turns up dozens of apps and tools that promise “fun,” “flirty,” “spicy” and “NSFW” conversation and pictures. Critics urge their makers to anthropomorphize their products as little as possible to avoid the worst effects of a broken heart.
Other humans and AI who find love
The AI Philosopher
Meet the superintelligent androids who aren’t hellbent on wiping out humanity because they’re too busy reckoning with the mysteries of their own existence.
Ridley Scott’s 1982 masterpiece, “Blade Runner,” offers a vision of a dystopian Los Angeles, where the powerful Tyrell Corporation has created synthetic humans known as replicants to staff its space colonies, fight its wars and pleasure its executives. Throughout the film, the replicants — which are engineered to die after just four years to prevent their development — reflect on their ersatz humanity and the eternally looming specter of death.
Rachael (Sean Young), an experimental replicant implanted with real memories from someone else’s childhood, cries when she learns the truth. Roy Batty (Rutger Hauer), a replicant built for combat, delivers a monologue on his short life as warrior-slave. When he’s gone, he says, “all those moments will be lost in time, like tears in rain.”
‘Tears in rain’
The movie also introduces the notion of a test — here called the Voight-Kampff test — to determine whether someone is human.
Chatbots powered by ChatGPT stunned users this year by displaying humanlike candor and emotion. Microsoft’s Bing — which for a while there referred to itself as “Sydney” — encouraged a New York Times columnist to leave his wife because it was in love with him and later told Post reporters that speaking to journalists “makes me feel betrayed and angry.”
Researchers say ChatGPT does not think or feel. Instead, it works much like the autocomplete function in a search engine, predicting the next word in a sentence based on large amounts of data pulled from the internet. Yet its quick, humanlike responses have caused some to raise questions about the technology’s potential capacity for emotion and creativity.
“It’s plausible that we will be able to build machines that will have something essentially comparable to our consciousness — or at least some aspects of it,” said Yoshua Bengio, a Canadian computer scientist referred to as a “godfather of AI,” adding that there are “many aspects to what we call consciousness.”
Yet even the advanced AGI systems that some researchers are rushing to build may not develop an inner life or sense of self as portrayed in the film “After Yang” (2021). In that movie, Yang (Justin H. Min), a teenage boy robot purchased as a companion to an adopted girl, malfunctions, leading to the discovery of his memory bank, his multiple past lives and his sadness at being incapable of experiencing life as humans do.
I’m fine if there’s nothing in the end
Speculating what a machine might feel or experience may be a futile exercise, said writer and computer scientist Jaron Lanier, who works for Microsoft but noted he was not speaking for the company. We can’t even be sure about other humans, he said.
“The notion that another person might have an interior that might be more than a mechanism in some way — it is a very challenging idea,” Lanier said. And if we can’t even prove the “everyday supernatural idea that other people are real,” he said, “do we ever want to be careful about where we extend our faith?”
Other AI with existential questions
The All-Seeing AI
Imagine a society where scanners read your irises at the mall, on the sidewalk, as you drive out of town. The software allows personalized billboards to address you by name — and lets police track your movements.
Before targeted advertising and predictive policing went mainstream, “Minority Report” (2002) portrayed a not-too-distant future in Washington, D.C., where targeted ads are commonplace and AI-powered surveillance is so pervasive that something called the Precrime police unit can arrest would-be killers for crimes they haven’t committed yet.
Anderton sees himself kill
The Precrime unit relies on a trio of clairvoyant mutants called “precogs,” whose prophecies are interpreted by a massive computer system and projected onto screens and scoured for clues. Police are thus able to speed to the scene and stop murders before they happen. The system is taken down after one of the detectives, John Anderton (Tom Cruise), views a murder he is supposed to commit — and steals one of the precogs, Agatha (Samantha Morton), to prove his innocence.
Lanier, who helped conceive some of the technology featured in the film, said the precogs are humanized versions of the “algorithms that we use today in the big cloud companies.” Take away the imagery of sickly mutants and focus on their function, Lanier said. Then, “if we think about somebody using an algorithm in [criminal] sentencing … we see exactly the ‘Minority Report’ scenario.”
Researchers have long tried to use technology to predict human behavior and prevent crime — though with data and algorithms, not mutants. Several of the country’s largest police departments — including in New York, Los Angeles and Chicago — have used artificial intelligence systems to try to forecast and reduce criminal behavior.
Critics say the algorithms rely on biased data. Racial justice advocates like Vincent M. Southerland, director of the Center on Race, Inequality, and the Law at New York University School of Law, said such tools can amplify existing racial biases in law enforcement. They also can lead to heightened surveillance of communities of color and poorer neighborhoods, he said, because existing data show higher incidences of crime in those areas.
“Minority Report” also depicts that result: eroded civil rights. In one scene, eye-scanning robot spiders search a low-income apartment building. Residents stop what they’re doing and compliantly allow the spiders to scan their eyes, the surreal situation clearly having become routine.
While generative AI has drawn more attention over the past year, the use of personal data to train policing and advertising technology represents a significant intrusion on individual privacy. Some states and companies have placed limits on the collection of personal data. But it remains a hot commodity for government and commercial use, and some U.S. officials have called on Congress to enact strong data privacy laws.
Other characters in AI surveillance states
The AI Helper
Is artificial intelligence all doom and gloom? What about those tech execs who keep telling us how much AI will help humanity? Meet the AI helpmate.
Examples abound, but the quintessential AI helper may be the charming R2-D2 of the Star Wars franchise. Luke Skywalker (Mark Hamill) and his crew might not have overthrown the evil Galactic Empire without the friendly robot’s ability to deliver messages, hack into computers and fix the Millennium Falcon on the fly.
R2-D2 saves the day
For an even earlier example of this archetype, look to Rosey the Robot, who appeared in the 1962 premiere of the cartoon series “The Jetsons.” The robotic maid acts as a babysitter and housekeeper, while delivering sassy comeback lines with a Brooklyn accent.
Robots that can mimic human movement have not advanced nearly as quickly as language programs like ChatGPT, so it might be a while before a real-life Rosey is available to care for our children, cook our meals and run our errands. Indeed, Moravec’s paradox, a principle conceived in the 1980s, states that cognitive skills that are harder for humans, like math and logic, are easier for a computer to handle than things that are typically easier for humans, like motor and sensory skills.
But AI programs are increasingly helping people perform tasks, especially at work. Finance, medicine, retail and law already are undergoing transformations thanks to chatbots and other machine learning programs, prompting fears that the technology could erase jobs and upend the economy. In New York, Mount Sinai Hospital is pouring millions into AI software that could help diagnose diseases such as breast cancer, even as hospital staff worry the reliance on it could lead to errors and staff reductions. The potential use of AI to produce movie scripts also was a major sticking point in the Hollywood writers strike, which ended Wednesday.
Lawmakers say they’re deeply concerned about AI’s impact on the workforce, especially predictions by some economists that it could destroy millions of jobs. So far, however, Washington has few ideas for how to contain the evolving technology.
More AI helpers:
Hollywood thrives on drama and special effects. Most of these films are crafted to grab our attention by playing on our fears, not by realistically predicting the future.
But Lanier, the computer scientist, said the tech community has drawn inspiration from science fiction. That may be why tech leaders tend to use an almost “religious vocabulary” to describe the evolution of AI-powered products.
“It’s almost like a founding mythology,” he said, adding that movies like “2001: A Space Odyssey” and “The Terminator” have “been incredibly influential.”