Tuesday, April 02, 2019

Artificial Intelligence is Everywhere (and has been for some time)

What might be "intelligence"?

One way of thinking of intelligence is the capacity to imagine options and their potential outcomes, evaluating those outcomes to make a choice. This definition raises many other questions: On what basis are the potential outcomes compared? How do we imagine options? Since there are a very large number of options, how do we decide when to stop imagining options? Suffice it to say that, to make choices, we make assumptions about the likelihood of options to yield desired outcomes, so that our search algorithms do not imagine options outside fairly short time and distance horizons.

What might be "artificial intelligence"?

What we teach each other is a form of artificial intelligence. This intelligence, based on another person's experience, saves us from having to rely on our own experience, letting us learn from previous generations. How do we know what to pass on to the next generation? Trial and error, that has given consistent results, is the most useful information to pass on. This especially includes learned lessons about cause and effect that contribute to our survival. Towards this end, we have developed the "scientific method" and the requirement that experiments must be able to be replicated across time and space. The scientific method is an algorithm for building artificial intelligence.

Culture, too, is a form of artificial intelligence. Culture permits us to share historical perspectives with each other efficiently and securely. Culture is about trust systems, like language, social etiquette, mores, belief systems, etc. Trust is critical to learning. If I don't trust what someone tells me, then I will expend energy testing and learning on my own. If I trust someone else's perspective, then I will more easily accept their perspective.

Why is artificial intelligence important?

As an individual, my view of the world is biased. First, because I have a specific existence in time and space. Second, because what I experience and remember as "reality" is biased towards (a) genetic (built-in) biases and (b) learned biases. The scientific method attempts to overcome the first and second biases by measuring and attempting to reduce the variability of the "experience", to make the variance dependent on the "experiment" and not the "experience". In other words, science is an attempt to make knowledge independent of any particular human being's experience of the world. This artificial intelligence is a more generalized and reliable explanation of reality, that exists in spite of an individual's specific experience of the world.