Cognitive Artificial Intelligence

The human beings' intrinsic ability to contemplate, understand, and communicate has perplexed cognitive scientists and theorists for centuries. In recent years, mankind has developed the tools and technologies necessary for simulating such abilities within computational programs. Many argue, however, that the true essence of cognitive thought has not been, and will never be artificially replicated. The primary basis for this argument lies in the nature of cognition; or the conditions necessary for an entity to be considered capable of thought. The modern belief that man will develop a machine which satisfies such conditions has lead many individuals to further contemplate the underlying principles of thought. For the purposes of this argument, I will consider the work of John R. Searle, and will present the factors I feel are necessary for cognitive thought to be realizable within artificial entities.

According to Searle, the creation of strong artificial intelligence, or rather, A.I. which equates or transcends human intelligence, is impossible. In his article, “Minds, Brains, and Programs,” Searle presents a number of assertions as to why strong A.I. is unattainable; the most notable being the Chinese Room argument. In this thought experiment, Searle demonstrates how a computational device lacks the ability to truly understand the logic behind its actions. It uses a series of rules in order to manipulate symbols, and the extent of its “thoughts” are no more than its ability to process a command according to its programing structure. Searle insists that, while a machine can be given the capacity to understand a particular item or subject, they are not themselves, capable of independent thought, (i.e., intentionality). In general, Searle's position on the cognitive possibilities of properly programmed computational devices is sound, and most of his supporting claims are well-founded. He does, however, lack a detailed account of the conditions necessary for artificially intelligent intentionality, and fails to address the many different forms of intelligence which exist.

The primary variable necessary for strong A.I., according to Searle, is understanding, or “the possession of mental (intentional) states.” However, his use of the word, “understanding,” is far too comprehensive. He is suggesting that an artificial intelligence must maintain the human mental states before it is considered capable of thought. While there is a definite requirement for a mental state, it is not necessary that they mirror those of the human being; e.g., beliefs or desires. In essence, an artificial intelligence needs only to indicate an ability to deviate between rational and irrational decisions before it can be considered a cognitive being. This ability, however, requires a few underlying essentials.

The first essential quality an artificial intelligence must acquire, before it can be viewed as a thinking entity, is the ability to demonstrate reason. This ability to make judgments is vital if the artificial intelligence is to exhibit any form of intelligence whatsoever. Within modern day computers and programs, this attribute has been somewhat established. The computers ability to defeat chess masters, calculate enormous equations and their subsequent solutions, and adhere to a countless series of if/then commands demonstrates, to an extent, the A.I.'s ability to reason. This, however, is only one of the conditions necessary to satisfy the intelligence question. The second quality is the ability to acquire new information. While some may contend an artificial intelligence cannot learn, it can be argued that learning, at its core, is no more then the absorption and retention of information. Through this definition, computers are already capable of learning at an increasingly fast rate, and thus, have satisfied the second requirement. Analogous to the ability to learn, however, is the notion of understanding.

As previously stated, Searle's definition and usage of understanding is not necessary for an artificial intelligence to be considered a thinking entity. What is necessary, however, is the A.I.'s ability to indicate the reason behind its reasoning. The third quality, to put it simply, is that the computer must be able to provide the logic behind its knowledge and decisions. This is perhaps the most important aspect of the intelligent entity, but is something which has yet to be developed within artificial intelligence. In the human model of cognition, it is vital to grasp both the fundamental and contextual aspects of a particular subject during the learning process. If one fails to do so, they are less likely to retain the knowledge; if one succeeds, they are more likely to understand the knowledge. However, in the computer model, the system has the ability to maintain information and knowledge indefinitely. For this reason, humans have largely consider computers to be nothing more then storage devices. If the computer has the ability to supply the reasoning behind its knowledge and resulting decisions, however, it can then be considered something greater.

Similar to a human being, the artificial intelligence can only exist as an intelligent entity if it maintains specific functionalities. For decades, mankind has imagined creating an intelligent being capable of reflecting the structure and rationale of human cognition. With the rise of computers and functional artificial intelligence, many individuals feel mankind is closer than ever to attaining this reality. However, with technological and scientific capabilities where they currently are, such a goal is highly unlikely. If mankind wishes to create another intelligent being, it need not pattern it off the totality of the human model. The metal state of the human being is simply too complex to replicate at this time. Instead, those in pursuit of such goals should focus on replicating the most simple and essential mental states which makes mankind conscious, and thus intelligent entities.