Artificial intelligence is a field of academic study with its own vocabulary and specialist terms. In order to understand the meaning of these terms, it is helpful to understand the history of artificial intelligence. A simplified graphic of this history is shown below;
Rather than a single line of incremental developments, the history of artificial intelligence is incredibly complex, resembling the repeated branching and merging of multiple streams (an anastomosis) of ideas. A more detailed (but still simplified graphic) is shown below;
In the simplified graphic we have divided the history of artificial intelligence (AI) into five broad periods;
- Origins (pre 1950's and early 1950's),
- GOFAI (early 1950's to the 1970's),
- Winter (1970's to the 19080's),
- Schism (late 19080's early 1990's),
- Resurgence (late 1990's onwards).
There are a number of key features in the history of AI that are sometimes overlooked from the main-stream Machine-Learning perspective, namely; the extent to which Cybernetics featured in the early development of AI and secondly, the schism that took place in the late 1980's and early 1990's.
It is perhaps not surprising that these two are overlooked, since not only are they somewhat related, but they also have philosophical underpinnings that are in opposition to those of mainstream AI. Namely, an organic perspective on AI, a perspective often accentuated by the use of the term adaptive behavior (journal).
Mainstream AI is primarily interested in the discovery of efficient algorithms for application in specialized problem domains. In contrast, adaptive behavior (AB) has at its core the notion that complex behavior is inherent in the activity of living organisms that have evolved to support their biological compulsions; namely that of survival and reproduction within their ecological niche.
The effect of this, is that whilst AI may be directly employed in the solution of a specific problem (for economic value or for scientific interest), AB is more likely to seek to reproduce complex behavior manifest in some living organism, and through this, indirectly solve some problem.
Obviously there are positive and negative aspects to each of these approaches. AI eschews the additional burden of mastering the biological sciences in its pursuit, whilst also freeing itself from the constraints of biological plausibility. However, the AB has something that AI does not, a bauplan for how intelligent systems are constructed as evident in their biological hosts.
Imagine a tortoise and a hare. The hare is fast, and runs ahead quickly exploring multiple pathways, some of which result in a dead-ends or cul-de-sacs. The tortoise however has in its position a map. Whilst the map may take time to decipher, the tortoise has some means of ensuring that it is heading in the right direction. The greater the complexity of the landscape, the harder it is for the hare to find the right path.
Artificial Intelligence (as the hare) has multiple goals, each one the solution to a specific problem. Adaptive Behavior (as the tortoise) has as its ultimate goal, the desire to simulate the complex behavior of intelligent organisms (including homo sapien), or even more fundamentally the desire to determine the general foundations and principles that underpin human-like (or greater than human-like) intelligence.
Our approach has its origins in the Adaptive Behavior approach. What this allows us to do, is arrive at a model that has extents encompassing the entirety of the brain.