What is Machine Intelligence?

By Tori Bowman

This is a question that basically boils down to semantics. For me to adequately answer what qualities a machine would need to posses for it to be considered intelligent, I first need to define what it means for something to be intelligent. Then, I would need to analyze which of those characteristics are essential and which are superfluous when applying that definition to a machine.

Of the several definitions that Webster’s Collegiate Dictionary offers for intelligence, a combination of two seems to be the best fit. By combining “(1) : the ability to learn or understand or to deal with new or trying situations; the skilled use of reason (2) : the ability to apply knowledge to manipulate one's environment or to think abstractly as measured by objective criteria (as tests),” a reasonable definition of intelligence can be achieved.

So, which of those characteristics in essential for a machine? For a machine, intelligence is the ability to learn, understand, reason, and apply knowledge. I do not feel that it is necessary for a machine to be able to manipulate its environment. There are some instances in which an intelligent machine can manipulate its environment, like a robot, for example, but that’s not a delimitating factor in determining intelligence. The same logic applies to the ability to think abstractly. While this could be a quality, it is not necessary for intelligence to be declared. I think that this particular quality would almost be unreasonable to expect from a machine. This would require a certain level of consciousness that is not required in my definition of intelligence.

At this point, I think it’s important that I settle what it means to be able to learn. Once again, I will consult Webster’s Collegiate Dictionary. Webster states that to learn means “to gain knowledge or understanding of or skill in by study, instruction, or experience.” In applying this to a machine, it would need to be able to gain knowledge or understanding about a subject or topic through studying, instruction, or experience. Most likely, a machine would use a combination of these methods. To begin with, it would need to undergo some instruction. After the initial instruction, it should be able to study situations and use its own experiences to gain further knowledge.

I like the point that the end of the second definition makes about how intelligence is measured. An objective criterion is the best way to determine if a machine meets the qualifications for intelligence. A machine should be able to be subjected to tests that require it to learn and then apply that knowledge to a later test. Also, there should be a test that would require the machine to complete a fair amount of reasoning. If a machine can successfully complete these tests, then it should be considered intelligent. Failing these tests would disqualify it from being considered intelligent. Also, if a machine possesses other impressive features, but lacks to pass these tests, would still be considered unintelligent.