There exist many definitions of artificial intelligence (Russell and Norvig, 2010), most of which unnecessarily focus specifically on machines. For example, New Oxford American Dictionary defines artificial intelligence as “the theory and development of computer systems able to perform tasks that normally require human intelligence, such as visual perception, speech recognition, decision-making, and translation between languages.” However, neither the term “artificial” nor the term “intelligence” suggests the involvement of a machine. New Oxford American Dictionary defines “artificial” as “made or produced by human beings rather than occurring naturally, typically as a copy of something natural” and “intelligence” as “the ability to acquire and apply knowledge and skills.” Therefore, artificial intelligence could be more generally defined as “the artificial ability to acquire and apply knowledge and skills.”
Artificiality of Intelligence
However, to confine the definition of artificial to man-made objects is also unnecessary. The term “artificial” stems from the Latin words “ars” and “facere” which mean a “skill in producing any material form, handicraft, trade, occupation, employment” and “to make (in all senses), to do, perform, accomplish, prepare, produce, bring to pass, cause, effect, create, commit, perpetrate, form, fashion, etc.”, respectively. Neither term suggests the quality of being man-made. Therefore, eliminating this restriction from the definition of artificial renders the definition of artificial intelligence as “the skillfully created ability to acquire and apply knowledge and skills.”
However, consider the possibility of a creation of a zygote of an artificially designed species of which a developed organism would be intelligent. Generally, a zygote is not considered to be intelligent. Instead, the zygote would theoretically develop into an organism and develop its intelligence itself via means provided by its genome. This begs the question of whether artificiality should describe the design process or the bearing agent.
Types of Intelligence
It turns out that it is helpful to focus on both aspects. To facilitate this, new terms must be coined. The first three terms, making up the designer class, are concerned with the designer of the intelligence and the latter three terms, making up the bearer class, are concerned with the bearer of the intelligence.
- Primary intelligence is intelligence that was designed by unintelligent agents.
- Secondary intelligence is intelligence that was designed by one or more intelligent agents.
Engineered intelligence is intelligence that was designed by a mix of intelligent agents and unintelligent agents.
- Mechanical intelligence is intelligence borne by a mechanical agent.
- Biological intelligence is intelligence borne by a biological agent.
- Biomechanical intelligence is intelligence borne by a biomechanical agent.
For the bearer class, the classification of the agent depends on both the storage items and the mechanisms through which the storage items are used since both are required to acquire and apply knowledge and skills.
Terms from different classes may be mixed. For examples, a member of Homo sapiens sapiens bears primary biological intelligence, the aforementioned organism would bear secondary biological intelligence, and Google’s AlphaGo bears secondary mechanical intelligence.
The current definitions of artificial intelligence suggest that the designer is man and that the bearer is a machine. This excludes many forms of intelligence that many would consider to be artificial. However, generalizing the definition of artificial intelligence reveals that the specification of artificiality would no longer be as meaningful. Instead, a classification system that identifies the designers and bearers of an intelligence is of greater utility.