What is intelligence?
Started getting into this question a couple of weeks ago, following Joanna Bryson’s presentation to the London Futurists – The future of robot ethics.
The Oxford defines it as:
“1 The faculty of understanding; intellect
2 Understanding as a quality of admitting of degree; spec. superior understanding; quickness of mental apprehension, sagacity.”
It seems clear to me that intelligence is more than simply “the ability to generate appropriate behaviour in response to an unpredictable environment” (Joanna Bryson’s definition), and certainly the need of individuals “to generate appropriate behaviour in response to an unpredictable environment” has been a key driver in the emergence of intelligence in biological systems.
Viewed from a systems perspective, there are some very simple systems that can produce appropriate behaviour in very specific circumstances. A thermostat can switch a heater on if the temperature drops below a certain point, and switch it back off once it reaches another point, maintaining a narrow range of temperature. In one sense, one can view biology as a relatively complex collection of such simple systems.
To get to more complex systems, you need quite complex sequences of such systems, hence the need to maintain very constant conditions in the womb as children develop.
It seems that the sort of intelligence that we experience in ourselves and each other requires the emergence of a very complex brain, and the emergence of a very complex language and culture, to be able to kick-start the sort of complex systems in language and memories that are capable of producing a model of the world to an emergent awareness in language, that is capable of being aware of itself (a software entity experiencing a software model of reality in the brain of a naked ape).
The ability to understand is based firmly in the concept of modelling.
Intelligence seems to be an ability to start with the models that evolution of bodies and evolution of cultures deliver to us, then to work with those starting points, and to further refine the models.
This seems able to be most powerfully done by questioning assumptions, and testing out new models of relating existing data, and by developing new tools that allow us to acquire new data, and to apply tests to that data.
In this way, our models evolve, and can become ever more useful at predicting and relating what we observe (or can become stuck in unexamined and inaccurate assumptions).
At each level of abstraction, the ability to model allows us to develop alternative possible futures, and to choose actions that tend to direct probabilities towards one or more of those possibilities, and away from others. And reality seems to be sufficiently complex that nothing we do has 100% guarantee of success, and some actions have much higher probabilities than others.
So it seems that intelligence has several major aspects to it:
Partly it is an ability to distinguish the patterns that exist;
Partly it is an ability to model variations of those patterns into possible futures;
Partly it is an ability to find patterns not yet existent, but possible, and to bring those into our models and then into reality;
Partly it is an ability to communicate with others whose patterns of understanding may be very different;
Partly it is an ability to find those actions that produce the greatest desirable change for the least possible action;
Partly it is an ability to shape one’s desires to be of greatest benefit to self, which ultimately means to all sapient life, not just some subset limited by family, village, or tribe or species or any aspect of culture (in its lesser sense) or some subset thereof.
It seems that intelligence is a very complex thing, that creates a great deal of diversity, and demands a great deal of acceptance and tolerance of such diversity.
The study of just how evolution by natural selection has managed to produce intelligence in human beings is fascinating, and has (and probably always will have) many gaps in the details, and the general themes are clear. Evolution, differential survival of variants in different conditions, leads to the emergence of some lines of ever greater complexity. All major advances in complexity are characterised by the emergence of new levels of cooperative behaviours with attendant stabilising strategies to prevent cheating.
Mastery of such awareness takes a very long time, awareness of the possibility of such mastery can come much quicker.