With the extraordinary expansion of technology in the 21st Century, new questions have arisen considering the ethics and implications of creating artificial intelligence as well as the extent to which machines are intelligent. With the advent of the computer, the whole world has changed. Many computers can easily process calculations far too complicated or time consuming for an average human being and help scientists solve some of the greatest mysteries in the world, yet the same machine could not answer the questions of a six year old child. An important question derived from this is: can a machine know? How is it that a machine can analyze the physics and properties of the cosmos millions of light years from Earth, or defeat the world’s chess champions, yet at the same time “know” so little?

One study by the British Government predicts that sometime within the next 50 years we may need to provide robots with “the same citizen’s rights as humans.” This prediction seems to imply that if machines, or robots, are not sentient yet, then they soon will be. Furthermore, what does knowledge imply? In order to know, do they have to be able to feel? In the popular science fiction film “iRobot”, humanity has invented many robots who know but are at the dawn of a new age as they have just invented the first one that feels. In our modern age, have we reached a point where we can create a machine that knows?

The American Heritage Dictionary defines “knowing” as, “possessing knowledge, information or understanding.” However this definition is glaringly incomplete in that it simply refers to other parts of the dictionary, which circularly refer back to knowing (for instance, knowledge is defined as “the state or fact of knowing”). Many pages could be dedicated to defining the word “knowledge” and still not be adequate, however I think it can be simplified. Knowledge contains many elements; it is a combination of knowing the facts, logic, and reasoning behind something, but it is also an underlying understanding of it.

Knowledge and understanding are intimately related in a way that it is hard to separate the two. While many machines can quite easily be defined as knowing the facts of something, some advanced machines may even have some sort of logic or reasoning behind their form of knowledge (even if their logic or reasoning is different than ours), but where all of man’s machines fail to meet the definition of knowledge is in their understanding of it. Understanding implies a thorough and in-depth comprehension of the topic, which children begin to grasp at a young age, but even the most powerful computers cannot manage.

The so-called knowledge that machines possess lies largely in the realm of computers. However, an understanding of how computers “think” exposes how their form of knowledge lies outside the boundaries of human knowledge. Computers function via a series of “if/then” statements. For example, simple chess computers function by looking at a move a player makes and then “knowing” that statistically the next best move for it to respond with would be X, based on the data inputted by the computer’s programmer. So the “if/then” statement would be: if the player moves here, then the computer should move here. Therefore, the best computer chess players are the ones with the largest databases of moves combined with the fastest and most powerful processing power, enabling it to cross check the move the player has made against every possible outcome in its database.

Chess is often regarded as a match of intellects, where the more organized and refined mind should win out, but the reason chess is hard for humans is that our memories our poor, we are easily distracted, and our ability to process what might occur multiple moves ahead is low. However, a good chess player is one who has outstanding knowledge and understanding of the underlying aspects of the game, so as to outwit his opponent. A computer, unable to compete with humans in knowledge or understanding of the game of chess, makes up for this gap through pure computational power. Chess enthusiast Ashley Dunn wrote that, “What it (the computer) was doing was analogous to trying to recreate Hamlet by typing, in random order, every word Shakespeare ever used … Admittedly … (the computer) had to exercise judgment in picking the best sounding words.”

Therefore, a modern computer is severely and fundamentally limited by its programming. It does not understand why the move it makes is the best move, it simply follows the rules of its programming. Computers may have found some answer to recalling facts, and show some logic through their programming, even a form of deductive reasoning in their processes and computations, however they have no understanding of the meaning or impact behind the results of their algorithms and computations, and therefore they lie firmly outside of the realm of knowing. While machines can be programmed with a vast amount of information and computational data, their lack of understanding and awareness of this knowledge inhibits them from knowing it. To obtain understanding it may require a level of self-awareness, or parallel thought processing and learning skills that machines have not yet obtained.

A computer program dubbed MEXICA made headlines as “the first computer to generate original stories based on computerized representations of emotions and tensions between characters.”3 Similarly to how a computer plays chess, at first it may appear that the computer “knows” how to write a book, but upon examination it is clear that the computer is substituting computational power and informational databases for understanding, thereby deserting it outside the realm of true knowledge. In this case, this innovative program generates a “rough draft” of a story from a database, which is extremely simplistic and may only be several lines in length. From this it designates its characters relationship’s emotional values on a range from -3 (hate) to +3 (love). Then using these numbers, it selects possible story elements (called atoms) from its database to create a short story with rising and falling levels of tension and intrigue, which it also measures quantitatively.

Finally, it cycles through the story again checking for “coherence and interestingness” before producing a final copy. When reading a story that this program produces it may seem that the computer understands the human emotions behind its characters, however by understanding how the computer creates these stories it is evident that its knowledge is founded in databases, its skill in innovative programming, and any form of understanding is substituted with mathematical algorithms. Though the program can create short stories, it does not understand what makes the story interesting beyond the limits of its data and programming. Because it does not understand it cannot know. A machine may fit an extremely simplistic definition of knowledge, but any more esoteric definition clearly excludes even the most modern computers.

The counter-point to a machine’s ability to “know,” hinges largely on the definition of knowledge. If a simplistic definition of knowledge is adopted, one in which knowledge is simple the ability to repeat back data, then the earliest computers, whose only form of user interface came in the shape of binary punch cards which required knowing technicians to then decode and decipher, could be considered “knowing.” If knowing is the ability to simply spit back data, then yes a computer knows. A machine can repeat back to you anything that can be inputted to it. The marvel of modern computing is their ability to process or compute immense amounts of data at the same time; combined with the massive amount of information now available through computers, or rather the vast amount of information we, as a species, have made available through computers via decades of programming. But is this truly knowing, in the abstruse sense of the word?

Machines are unable to know because they are unable to understand. They are unable to understand because at their heart they are simply complex machines for repeating back data and have no sense of comprehension or awareness for the data they hold. They may be able to repeat facts, emulate logic, express deductive reasoning, but they gravely lack understanding, which they hopelessly try to make up through computational power and processing. However, nothing yet discovered or innovated can emulate the inquisitive and self-aware psyche that makes the human mind unique. Human thought is a product of millions of years of evolution and it is a pity to think that a few meager decades of programming would result in the second coming of the mind in the realm of machines. That said, the breadth and scope of the human mind will inevitably provide for the innovation of true knowing, thinking, and feeling machines sometime in the-perhaps-not to distant future.