Theory
What are the elements of a "human"?
Artificial Intellegence
One of the first elements that is needed to create a Virtual Human is some level of "Artificial Intelligence". Now AI is a bit of a hot topic right now but lets talk about the actual definition, which is much broader than one might think. Basically, Artificial Intelligence is exactly what is sounds like; a level of intelligence that artificial agents (i.e. virtual humans) exhibit in contrast to the natural intelligence that humans exhibit. Just like natural intelligence, these systems take in information and make decisions about what actions to take. This can be anything from your Rhumba deciding whether to turn left or right to the automatic doors at the grocery store knowing when to open to complex intelligent chat-bots that we are going to discuss further here. Adding some level of information processing and decision making is the backbone to all virtual humans. [10] However simply having a machine make decisions does not make it a full virtual human. Just like real humans are not simply decision making systems, we need more elements to a system in order to consider it to be a full virtual character.
Natural Language Processing
Once a system can take in information and make decisions, one of the next most important aspects for creating a virtual character is allowing that character to communicate in a way most similar to humans. Communication is the center-point of interaction so in order for a virtual human to feel like a real character, it must have a level of communication in order for people to interact with it. Primitive forms of communication are simply having a prerecorded voice read text written by humans. Clifford Nass showed that this is often enough to produce a social response from humans when interacting. [11] However text generated by a computer is more new. Natural Language Processing (NLP) is the study of how computers can communicate to humans most naturally. In the beginning, most NLP systems were based on complex sets of hand-written rules, based on the theory of Transformational Grammar [12] and Chomskyan Linguistics. Chat-bots like ELIZA [3] could produce a general response to basic conversation using these models. Once these theories started being replaced by the theory of Corpus Linguistics [13], higher levels of communication for computers became available through a specific form of Artificial Intelligence called Machine Learning. With this new technique, computers could more accurately respond to anything that is said to it. Advancements in Natural Language Processing allowed systems like Siri and other virtual assistants to come into the consumer market.
Social Psychology
The easiest way to tell if something is human or not, is to simply interact with it. Many publicly-understood rules govern these interactions and the study of such rules is the field of study for social psychologists. In general, these social psychologists want to understand why individuals make the choices, and have the reactions, they do when placed in social situations [14]. Most of the research in the past has been about how humans interact with other humans, however as we have seen, these social situations can now be defined by interactions with virtual humans as well. What social rules carry over into interactions with virtual characters? This question is still an open one being dissected all over the world. Some things have already been defined though. For instance, it has been proved that humans apply personality traits to computer systems [15] and more. Do you think you would treat a virtual character the same way you would treat a human? If it looked and acted like a human? Whether or not you fully realize it, there are certain rules defined in these relationships that are still being discovered today.
