The Internet is abuzz with discussions about whether artificial intelligence can actually become aware after a Google chatbot reportedly stated it was just that.
But while the internet may now be freaking out about AI sense, Hollywood has been obsessed with the subject for decades, specifically exploring the idea through robot-human clashes in movies like I robot and the terminator franchisee.
Perhaps the most recent version of this controversial AI debate is in: Archive, a 2020 science fiction film streaming on Amazon Prime. The film delves into a range of sci-fi topics, including uploading your consciousness to a digital afterlife, but its central premise teases one particular question: Can we create robots that are truly “equivalent” to humans?
inverse spoke to two experts in robotics and artificial intelligence to unravel the heavy science at the heart of this stunningly dark sci-fi film and ask if it’s mischaracterizing AI in its quest for human equality.
“There are many different definitions of human-level AI”, Baobao Zhanga assistant professor of political science at Syracuse University who researches AI governance, tells inverse†
Reel Science is a inverse series that reveals the real (and fake) science behind your favorite movies and TV.
Will AI become “equivalent” to humans?
ArchiveThe protagonist, an isolated scientist named George Almore, has built three different Android prototypes – each more advanced than the last – with the aim of creating an AI that is truly “equivalent to a human.” An Android is a robot that looks like a human. He tests the ability of his prototypes to display inherently human qualities, such as empathy, through a video game in which he plays with a puppy.
“You are all attempting the same thing. Deep learning. Artificial intelligence. Human equivalent of the holy grail,” George tells his third and most advanced prototype, J3.
But if you ask an AI researcher if humans can build an “equivalent” robot — or perhaps a robot that surpasses humans — they’ll give you a more complicated answer than the movie offers.
“I think it’s important to unravel what AI surpassing humans means. We need to distinguish between AI that outperforms humans in specific tasks and AI that outperforms humans in all or almost all tasks,” says Zhang.
For example, Zhang’s research focuses more on defining what she calls human-level AI in relation to work.
The trailer for the 2020 sci-fi movie, Archive†
“In our study, we defined ‘human-level machine intelligence’ as when machines can perform 90 percent or more tasks better than the average human worker paid for that task,” explains Zhang.
Sven Nyholmassistant professor of philosophy at Utrecht University and author of the book Humans and robots: ethics, agency and anthropomorphism† feels the same as Zhang.
“Well, it would be good to know the answer to this question: in what way equivalent?” asks Nyholm.
Nyholm says it’s plausible that humans could develop AI that behaves in the same way – in other words, mimics – human behavior in a ‘narrow range of situations’. But a robot that functions on the same level as humans in all situations seems much less realistic.
Further, speaking of an AI that matches not only humans in cognitive intelligence, but our full emotional range and capacity for empathy, that’s probably more unlikely, despite what Archive suggests.
“If we have an AI technology that doesn’t have an animal or human-like brain or nervous system, it’s hard to see how it can feel feelings or have affective states similar to ours or animals,” Nyholm says.
But there is another less obvious definition of “equivalence”: a moral one. Nyholm describes the research of John Danaher, which states that if a robot behaves in a manner equivalent to that of a human, it must have the same moral status as humans. Nyholm is a little less certain.
“But of course that’s a big ‘if’, since it’s very difficult to make robots that behave the same way humans behave,” Nyholm says.
Can we really compare AI to humans?
In ArchiveGeorge uses a lot of vaguely scientific mumbo jumbo to explain how he built his prototypes, throwing around real terms like “deep learning” — a method of AI training modeled after the way humans learn — but is one of are film science rooted in actual AI research? Not really.
Research into deep learning in practice is doing model AI stats against human performance, according to Zhang, but that’s about all the movie does well. It’s in the details of the movie where Archive leaves the realm of real AI research and enters science fiction.
The film’s scientist, George, has developed three different versions of androids, one more advanced than the other. The first prototype stopped mental development at the age of five and is emotionally monotonous. The second prototype is mentally more developed than the first and expresses basic emotions such as jealousy. The third and final prototype is believed to be “equivalent” to a human and contains all the complexities of being human. He uses equivalent brain scans of people of different ages to show how far each prototype has progressed.
But real experts say this is too simplistic a comparison between AI and human development.
“The idea that we can link AI development to human development — like, say, in a certain AI system that’s equivalent to a five-year-old — is pretty unrealistic,” Nyholm says.
The reason it’s unrealistic has to do with AI systems that work in a very different way to human bodies. While AI can get really good at specific tasks such as: to play the game To go, it is often very bad at other tasks outside its area of expertise.
Zhang agrees. “I think it’s hard to map AI development to human development,” she says.
Do AI researchers care about “equivalence” as much as Hollywood does?
Despite all the attention Hollywood has paid to movies investigating robots that have rivaled or surpassed humans, it’s not much of a concern for researchers working in this space.
“I wouldn’t say that’s a big topic of discussion among AI researchers. Most AI researchers work on much smaller topics,” Nyholm says, but he adds that researchers occasionally consider science fiction scenarios to inspire their work.
Researchers to be interested in making sure AI doesn’t harm humans, but not because they’re worried that robots will become too intelligent, as Hollywood likes to suggest. An example could be racial prejudice in facial recognition technology, which relies on computer-generated algorithms.
“I think it’s very important that ‘narrow’ AI systems deployed today are secure, fair and robust,” Zhang says.
Zhang adds that Hollywood movies like Archive tend to focus on plots that anthropomorphize — ascribe human qualities — to robots, which can lead the general public to misunderstand how AI works in our everyday lives in software applications such as chatbots or search engines.
“Most AI systems today are not androids or robots; instead, they are embedded in software applications without a physical representation,” says Zhang.
“I think it probably doesn’t really match a lot of real-world AI research. However, it makes for a good and compelling story in a science fiction story,” Nyholm added.
Archive is now streaming on Amazon Prime.