Cognitive AI and Engineering AI
[The] main methodology [in Artificial Intelligence (AI)] is the exploration of cognitive theory by building intelligent artifacts. Though the design of any intelligent artifact would be classified as an AI, AI as a discipline is united in the core belief that intelligence is a kind of computation. Thus, in practice, AI artifacts are almost always computers or computer programs. This also explains why AI laboratories typically are found in computer science departments.
Engineering AI is concerned with how to design the smartest intelligent artifacts possible, regardless of whether the processes implemented reflect those found in natural intelligences. The vast majority of AI research falls into this category. Cognitive AI, in contrast, endeavors to design artifacts that think the way people (or sometimes other animals) do. A subcategory of cognitive AI is cognitive modeling, which tries to quantitatively model empirical human participant data. Many cognitive modeling groups are working in psychology departments. AI cognitive models are always implements information-processing theories. That is, the theory describes intelligence in terms of content, representation, access, use and acquisition of information, as opposed to, for example, a statistical model of the influences on IQ (e.g., age) in a population.
The original dream of AI was to develop human-like intelligence [...].
-- Ashok K. Goel , Jim Davies
from "The Cambridge Handbook of Intelligence"
Quoted on Sun Nov 4th, 2012