Where will mankind be in 2050? Out colonizing Mars? Perhaps collectively uploaded to an artificial neural network? According to a semi-intelligent android, we may well be the subjects of his people zoo.
The android – created to resemble the famed science fiction writer Philip K. Dick – is the brainchild of roboticists and modern day Frankensteins, Andrew Olney, and David Hanson of Hanson Robotics Ltd. “By resurrecting PKD as an android,” David Hanson writes, “we seek to realize genius-level AI with compassion and creativity.”
(See our interview with Creative Director Stephan Vladimir Bugaj here.)
Though the PKDroid’s (our name, not there’s) physical likenesses to PKD are imperfect, it seems to dream up similarly fantastic futures and bleak scenes for mankind. This is in large part due to the database its creators provided.
To replicate the author’s intellect as close as possible, Hanson and Olney uploaded almost 20 PKD novels, plus speeches and conversations with the author onto PKDroid’s software. This enabled it to respond to inquiries in a similar fashion to it’s human prototype.
Pre-programmed replies, including interviews with other writers and integration with the Internet, gave PKDroid enough grit to hold an apparently intelligent conversation with a probing journalist. If it turned out that the same question was asked to PKDroid and PKD during his lifetime, the former would respond in line with the latter. Meanwhile, PKDroid answered unexperienced and unexpected questions with help from a process called latent semantic analysis, a mathematical technique that enables a program to interpret meaning from natural language.
On top of its linguistic capabilities, PKDroid is capable of identifying individuals and recognizing their behavior through biometric-identification software and advanced machine vision. Hanson Robotics notes that the robot can “track faces, perceive facial expressions, and recognize people from the crowd after a limited period of interaction.”
The PBS Nova reporter challenges PKDroid with questions most freshman philosophy majors would even stutter to answer. “Do you agree with Descartes,” he asks, “in ‘I think therefore I am’? Do you think?” PKDroid calmly evades the question, and instead suggests that the interviewer’s assumption is that all of the android replies are programmed. “The best way I can respond to that,” he says, “is to say that everything, humans, animals and robots do is programmed to a degree.”
PKDroid isn’t without moments of severe awkwardness, such as when he describes his ability to learn from grammatical mistakes, integrate new information and make daily progress in conversation. “Pretty remarkable, huh?” he asks, but his ‘huh’ sounds more like a gasp of surprise than self-affirmation.