closed, their activity lights dimmed. Nothing moved, nothing stirred. It was so still that Egerton could hear the blood coursing through his ears.
* * *
“Well, then I guess we aren’t understanding each other then because it sounds like you’re tell me that our mining bots now somehow believe in God.” XienCheng’s [the mine’s administrator] jowls shook with frustration.
“It’s not that,” Egerton tried again. “It’s not God. They don’t believe in God. They believe in going to church, in the action of going to church.”
“But that doesn’t make any sense,” XienCheng interrupted.
“Yes!” Egerton slammed his hand down on the table. “Now you’re getting it.”
XienCheng drew back in an awkward fear. “No. No, I’m not.”
“They aren’t doing anything in the church,” Egerton explained. “They aren’t worshiping God or saying prayers or holding a service; they’re just going to church. It’s the action and it’s not supposed to make sense. That’s the point. For some reason they latched onto the idea but now they need it. It keeps them safe.” (Johnson, 2009)
Mystery solved! The robots at the Piazzi Mine were going to church precisely because it was irrational. Like some human behavior, irrational actions can be as constructive as rational decisions. Most people think that robots and artificial intelligences (AI) can only make rational decisions. But we humans make good decisions and bad decisions, and it is because of this that we learn much faster than if we only made “correct” or rational decisions. It is this complexity of learning that allows humans to operate in highly complex environments.
Now you might be saying: “But that’s just science fiction!”
But you would be wrong …
The following abstract is from a paper with the impressive title: Using Multiple Personas in Service Robots to Improve Exploration Strategies When Mapping New Environments. This research was the work of three scientists Dr. Simon Egerton, Dr. Victor Callaghan, and Dr. Graham. [Note: You may have noticed that the name of the main character in Nebulous Mechanisms and one of the authors of the research paper are the same: Dr. Simon Egerton. This was originally done as a homage to Dr. Egerton as he was the first person to explain to me the work that the team was doing over a pint of beer in Seattle Washington. When I named the character none of us realized that the SF prototypes would be so popular and effective, spawning a whole series stories. More about that in Chapter 7. If it is any conciliation—the main character of the stories pronounces his name “Egg-er-ton” whereas the good doctor pronounces his name “Edge-er-ton.” I am indebted to Dr. Egerton (“Edge-er-ton”) not only for is mind-expanding research but also for being so understanding and having an excellent sense of humor about me dragging his name all over the world.] Published in 2008, the paper explains the emerging research by the scientists who were using a concept from philosophy and psychoanalysis in their robotics and AI development work.
Abstract
In this paper we describe our work on geometrical and perception models for service robots that will support people living in future Digital Homes. We present a model that captures descriptions of both the physical and perceptual environment space. We present a summary of experimental results that show that these models are well suited to support service robot navigation in complex domestic worlds, such as digital homes.
Finally, by way of introducing some of our current, but unpublished, research we present some ideas from philosophy and psychoanalytic studies which we use to speculate on the possibility of extending this model to include representations of persistent experiences in the form of multiple personas which we hypothesize might be applied to improve the performance of services robots by providing a mechanism to vary the balance of current and past experiences in control decisions which appear to serve people well. (Egerton, Callaghan and Clarke, 2008)
The idea of multiple personalities is not a new concept to people. The concept was brought to the attention of popular culture with the 1973 book Sybil by Flora Rheta Schreiber. The book was about a patient, Shirley Ardell Mason, who was being treated for dissociative identity disorder by her about her psychoanalyst Cornelia B. Wilbur. Mason’s condition was more popularly known as multiple personality disorder. In 1976, a popular made-for-television mini-series was produced based on the book starring Sally Field as the patient and Joanne Woodward in the role of Sybil’s psychiatrist. Field won an Emmy award for her performance. The Internet Movie Database describes the movie like this: “The true story of a young woman named Sybil, whose childhood was so harrowing to her that she developed at least 13 different personalities” (IMDB, 2010).
Egerton, Callaghan, and Clarke take the multiple personalities and apply it to the development of robots and AI. They expand multiple personalities into multiple personas, which has a little bit of an expanded meaning. Most think of a persona as the role that we as people play in society and in our lives, whereas a personality is a collection of traits that make up a person. The highly influential Swiss psychiatrist Carl Jung defines personas as “the mask or appearance one presents to the world.” Egerton, Callaghan, and Clarke use this idea to radically rethink the design of AI and software systems.
Multiple personas in philosophy and psychoanalysis are seen as one explanation for irrationality and are based upon the generally pathological process of splitting. In this instance splitting for functionally sound reasons is suggested as a possible aide to robust and efficient working within a variety of different contexts. In this sense it is more akin to the benign forms of splitting that allow us to be parents, siblings, children, workers, partners etc. (Egerton, Callaghan and Clarke, 2008)
The scientists are using “multiple personas” in a constructive and pragmatic way in the development of their robot’s AI. The splitting of the personalities in this case is not negative as with Sybil. They take a very different approach. They view the splitting of the personalities into personas as a highly efficient and functional way to segment a robot’s intelligence or brain. By separating the functions of the robot into these personas, it becomes easier for the scientists to design and develop the robot’s intelligence.
This persona based approach to the architecture will allow us to explore the hypothesis that multiple personas guide our actions, that we do not make decisions purely on our immediate sensing of the world. By having a reservoir of specialised personas to call upon, the persistent and evolving nature of such personas would allow us to explore the value of accumulated experience that in us manifests itself as a somewhat ill defined ‘self’; which, when making decisions, occasionally overrides the logical nature of the world, akin to what might appear to be irrationality, putting it down to nebulous mechanisms such as “a hunch” or “a feeling.” This approach will allow us to open up a line of research to explore the nature and value of such abstracted personas and their dynamics. (Egerton, Callaghan and Clarke, 2008)
Nebulous Mechanisms acts as an SF prototype for the scientists’ vision for an approach to programming AI. The story allows us to imagine some extreme effects of their approach and what it might mean to have “irrational” robots. What could be the positive and negative effects of this? Ultimately,
