The purpose of this post is to expose executives and marketers to changes in the world of AI and explain why these developments are relevant for a wide variety of businesses and industries.
AI and Our Everyday Life
I get asked a lot about what is artificial intelligence. There's a lot of terminology out there like:
There is a lot of hype around these terms, but not a lot of lucid and clear information about what any of it really means.
People often times make AI far more complicated than it actually is. Many people's vision of AI is shaped by Hollywood - think The Matrix or Terminator. Others are shaped by the ultra-high-tech movement which generates a lot of buzz these days - think self-driving cars, Google's DeepMind, and IBM's Watson.
Some of those are wonderful examples, but in reality, most AI tends to hide in plain sight.
Healthcare diagnostics, for instance, and the algorithms that power MRI machines, cardiac pacemakers, and even insulin pumps, are all powered by adaptive algorithms.
Telecommunications including call routing, internet search engines, and Apple's siri, all have well developed machine learning back ends associated with them.
Even less sexy things like developing infrastructure, utility monitoring through anomaly detection, and analysis of power grid network structure apply the ideas of graph theory and machine learning all at the same time.
And of course the early precursors to AI and machine learning have been used in finance for decades. Some of the most legendary financial institutions have used AI for forecasting and optimiztion and would not be where they are today without machine learning.
What is Artificial Intelligence?
So what is AI? In my opinion, AI is a measuring stick with a purpose. We are using a mathematical yard stick to measure the distance between one set of observations and another set of observations to solve problems based on the distances.
A good example that I like to use is this:
Think about apples and oranges. How does one mathematically predict based on some measurements about the fruits, which is which?
If I gave you twenty observations of the color of 20 randomly sampled apples and oranges, what would we see? R-R-R-O-O-O-R-R-R-O-O-O etc. Most of us could probably predict which is which - apples are red and oranges are orange!
Now, how do you get a computer to do this from something more complicated, such as the millions of pixels in a sample of digital photos?
How Does AI Work?
Here comes the yardstick. Since a computer cannot "see" color, shape, or any other feature that comes naturally to humans, we need a way to measure distances between observations.
Observations that are "closer" are probably the same fruit, and past a certain distance we might come to believe that anything past a distance threshold belongs to the other class of fruit.
We are finding ways to represent data, measurements, and events in 1's and 0's so that our algorithms can learn to measure the distance between observations and most crucially, be able to predict classifications based on these distances.
AI is a mathematical yardstick applied purposefully to problems. Much of the art in practice comes in being able to transform data and problems into representations computers can understand and measure.