In our previous post, we considered the Philip K. Dick novel Do Androids Dream of Electric Sheep? in the context of a thought experiment: a thinking tool used by scientists and philosophers to reframe an unintuitive problem in a new, more familiar context.
Another great thought experiment of the 1960s is 2001: A Space Odyssey. Developed in parallel as both a novel by Arthur C. Clarke and a film by Stanley Kubrick, Space Odyssey is a story about the evolution of humanity and intelligence. Beginning with the invention of tool use in Africa, the story asks how intelligence might continue to evolve after humans develop sophisticated technology. Both the novel and the film follow astronaut Dave Bowman on his journey to the Outer Solar System guided by HAL 9000, an artificial intelligence (AI) that controls the spaceship and is responsible for the well being of the crew.
“HAL is indeed Hofstadter’s vision of frighteningly competent AI.”
While embodied inside a computer console, HAL speaks in a human voice with an eerie calmness, almost like today’s affectless text-to-speech software, but with a colder, more calculating personality. HAL is capable of abstract reasoning, language comprehension, language production, recognizing human faces and emotions, singing, and admiring Dave’s drawings.
A symbolic scene in the film occurs when HAL beats Dave at a game of chess, shortly before HAL’s maliciously deceives and betrays the ship’s crew out of distrust. HAL’s chess victory on the big screen foreshadowed the 1996 victory of IBM’s chess AI computer Deep Blue over world chess champion Garry Kasparov. In 2017, we take computer superiority at chess for granted. Yet cognitive scientists prior to Deep Blue were skeptical that AI would achieve such a milestone before passing a Turing test, the benchmark for computer intelligence discussed in last week’s post. As cognitive science Douglas Hofstadter wrote in his 1979 book Gödel, Escher, Bach:
“There may [eventually] be programs that can beat anyone at chess, but they will not be exclusively chess programs. They will be programs of general intelligence, and they will be just as tempermental as people. ‘Do you want to play chess?‘ ‘No, I’m bored with chess. Let’s talk about poetry.‘”
HAL is indeed Hofstadter’s vision of frighteningly competent AI. Though today’s AI still struggle with understanding language and processing human emotions, serious thinker such as Stephen Hawking and Elon Musk have questioned whether AI might not one day betray humanity, just as HAL betrays Dave and his fellow astronauts.
“Clarke seems to imagine machines as a stepping stone in evolution …”
Indeed, such a day of reckoning might not be far in the future. Recent advances in AI have been achieved using artificial neural networks, computer programs inspired by the architecture of the real brain. Rather than having preprogrammed knowledge or output, these networks learn in much the same way that the brain learns. Connections between neural units that simulate neurons are altered according to Hebb’s rule: neurons that fire together wire together. In other words, associations between stimuli in the environment are captured as associations between neurons in a network. Artificial neural networks are already being used by Facebook to understand conversations and automatically tag photos of your friends. DeepFace, a neural network currently used by Facebook, recognizes faces with 97% accuracy and has 120 million neuronal connections whose strengths change with learning (by comparison, the human brain has 100 trillion neural connections).
One can imagine HAL 9000 as a super intelligent neural network following in the footsteps of DeepFace. After HAL’s betrayal in the film and novel, Dave embarks on a metaphysical journey through the next stages of human evolution. At the heart of Space Odyssey is the question, “What is the next step in human evolution and what role will machines play?” Ultimately, Clarke seems to imagine machines as a stepping stone in evolution between biological forms of life such as ourselves and an ethereal form of life symbolized by the mysterious Star Child appearing at the story’s conclusion.
Not all works of science fiction involve androids or outer space. In the next Knowing Neurons’ post, we focus on another classic of science fiction, Daniel Keyes’ Flowers for Algernon. Keyes’ novel asks questions that are all the more meaningful when considered in the wake of AI and Space Odyssey’s HAL 9000: What is the true value of intelligence? What are the ethical implications of artificially manipulating a person’s intelligence? Find out next week!
Gillings, M. R., Hilbert, M., & Kemp, D. J. (2016). Information in the Biosphere: Biological and Digital Worlds. Trends in ecology & evolution, 31(3), 180-189.
Hofstadter, D. R. (1979). Gödel, Escher, Bach: An eternal golden braid. New York: Basic Books.
Simonite, T. (2014). “Facebook Creates Software That Matches Faces Almost as Well as You Do.” MIT Technology Review.