We are going to touch on a ton of cool stuff in this post but here is the big idea: fundamental beliefs can keep us from understanding artificial intelligence.
Now what the heck do I mean by that? What i mean is that artificial intelligence, both as an idea and potentially as a product, spans a massive idealogical spectrum from engineering and computer science all the way to philosophy. And so to understand what AI is and what AI means requires reconciling the idea of artificial intelligence with beliefs all along this spectrum. Mainstream media consistently fails to do this, and so the most widely read material on the topic of AI is fundamentally misguided. So, we are going to take a look at where things often go wrong….
Fundamental belief 1: intelligence is an exclusively human quality.
This is a tough one, because it runs deep. In some sense, modern scientific history (last 2,000 years) can be described as man’s slow realization of his own insignificance. At the core of this evolution in perspective is two competing belief systems: scientific truth and Judeo-Christian values. On the religious side we find deeply held views about human exceptionalism. According to most western religious texts, we owe our best qualities (intelligence, consciousness, etc.) to the will of god, who carefully chose to give people these advantages but not the other creatures of the world. But in the 16th century, these views were challenged by Charles Darwin’s theory of evolution. All of a sudden, there is an alternative explanation for the origin of man, and therefor also an alternative explanation for the exceptionalism of man. It made us think: if we evolved by the same processes as other animals, how exceptional are we really? Once we started seeing ourselves as animals (or at least as related to animals) it opened the door to a much deeper understanding of ourselves. It also opened the door to a much deeper understanding of our own intelligence. Intelligence became relative. We began to realize that intelligence isn’t a narrowly defined ability that just humans have, rather intelligence is a fluid, relative concept that appears all over the natural world. It is a relatively knew idea to recognize intelligence in all of the many shapes and forms it comes in – especially in nature. But its critical idea for understanding AI. Because in order to understand artificial intelligence, we must first broaden our definition of intelligence itself.
Fundamental belief 2: computers will never be able to do x.
I will concede that there may be some things computers cannot do – for arguments sake. However the point is that when most people draw this line (the line between what is possible for computers vs. what is impossible for computers), its arbitrary. And also, its a line that has constantly shifted over the last 100 years. So, if we can clearly see that we have been underestimating the abilities of computers, then why do we keep on doing it? The answer is simple: once we understand exactly how something works, it ceases being impressive to us. It was said for a long time that in order for something to do calculus, it would have to be intelligent. Yet it wasn’t long after computers were invented that computers could do calculus perfectly. It turns out calculus can be strictly defined by a set of logic (code). But once this program was written, people didn’t say “wow, how intelligent is the computer running this program!”. Rather they said “wow, how simple is calculus!”. Just because we don’t always intuitively understand how a process might be defined in a way that a computer can understand – doesn’t mean its not possible. Its an important point for understanding AI, and it brings us to our next one.
Fundamental belief 3: computers are stupid because they can’t do things that are easy for humans.
Some people might say “If robots are really that smart, why can’t they do things that are easy for me, like carry a tray of food in a restaurant? Why do we still have waiters?”. The simple answer is that balancing a tray of food is actually a tremendous feat of intelligence. It may not be intelligence in the way we normally think about it, like the kind of intelligence you use to solve complex problems. But its still a massive amount of information flowing all throughout your nervous system in cycles of continuous feedback. There are over 600 muscles in the body, and they all have to coordinate to pull off this feat. Another example of this is driving a car. For people, its a routine task that can be done while having a conversation or eating a cheeseburger. But computers are only just now getting the hang of it. Why? Because driving a car is actually extremely difficult. The amount of visual information that we constantly process is staggering. Difficulty is relative, and just because a task is easy for a human doesn’t mean its easy in absolute terms.
These are just a few of many fundamental beliefs that prove as obstacles in understanding artificial intelligence from the proper perspective. These beliefs can be wide ranging, and they tend to be supported by deep, long held, and potentially disproven ideas about reality. That is what makes them so obstructive. Artificial Intelligence is really a set of many ideas, and to understand AI as a whole is to understand how its underlying ideas support each other.