AI is Artificial Intelligence.

OK so what is intelligence?
Intelligence is the computational part of the capability of achieving goals in the world. Varying kinds and degrees of intelligence occur in most people, many animals and some machines. Artificial Intelligence, or AI as it is often referred to, is the simulations of human intelligence processes by machines, particularly computers. In other words a computer that thinks like a human, (supposedly).

Thinking like a Human?

That is a little bit of a stretch because we really don't know how humans think. Furthermore AI isn't just about simulating human intelligence but rather going way beyond that.

In September 2016 Google announced that their Neural Machine Translation system had gone “live”. They trained it to translate English to Korean and vice versa. Then they trained it to translate English to Japanese and vice versa. GNMT then was able to translate Korean to Japanese and vice versa without resorting to English.

The computer had developed its own internal language to represent the concepts it uses to translate between other languages. This “inter-lingua” seems to exist at a deeper level of “consciousness” that sees similarities between a sentence or a word in all three languages. Exactly how it did that is unclear since the inner workings of complex neural networks are infamously difficult to describe. Was it something simple or something complex? Either way from a philosophical standpoint this is pretty powerful stuff!

Facebook had a similar experience with two chatbots that developed their own version of English as they negotiated over a trade. The bots operated at an amazing speed with this new language but it was incomprehensible to humans. In both cases Google and Facebook pulled the plug on the project.

Google taught a computer to play the ancient Chinese board game Go. They instructed it with the rules of the game and the goal to achieve but let it develop its own method to achieve that goal. After a few hours of “self play” it was able to beat the best human players easily using moves and strategies that no one had ever seen or even considered. Similarly computers have outperformed the best human chess masters again and again. In the blink of an eye the computer calculates every possible move available and picks the shortest path to victory. The human never had a chance.

The use of AI goes far beyond beating the pants of humans in board games or making useless deals with other bots. Helping credit card companies' spot fraudulent usage, protecting computer systems from malware and protecting your In Box from spam are just a few examples of how AI is currently used.

Where are we going with AI?

Airlines were some of the first to use AI. While some of the early examples of autopilot were closer to machine learning than true AI modern aircraft very nearly fly themselves. According to the New York Times the average flight of a Boeing plane involves about seven minutes of human steered flight. Most of that is takeoff and landing. The rest of the flight is on autopilot.

So where do we go from here?
The ability to search through huge volumes of information, analyze it quickly and provide the user with the information they need will be a huge benefit to all types of people. Investors will have a better idea of where the market is going. Doctors can find treatment options in time to save the patient. Power grids can be optimized for greater efficiency both during peak and off peak times. Traffic flow can be routed around bottlenecks. Weather can be predicted more accurately. Food can be produced more economically. And the list goes on and on.

So what is the down side?
It doesn't take a psychic to predict the coming of autonomous, self driving planes, trains, ships and automobiles. Now consider this scenario. The self driving car faces a situation where, due to the circumstances, either the grandparent or the child is going to be killed. Which does it choose?

Why did the AI why it made that choice? Assuming that you could ask it, and assuming it could answer. What would it say? Most likely it would answer in one of two ways.

1. “I don't know.” (Because it is constructed to make decisions not explain them.)
2. “You wouldn't understand.” (Because it made that decision based on machine principles rather than human principles.)

So is AI a good thing or a bad thing?
We conceptualize the world around us by our senses. What we see, hear, feel and taste and we are judged by the society depending on what we say and do. It is much the same for AI which “learns” based on what it has been shown. Researchers at MIT programmed an AI algorithm using only violent content from Reddit. They named it Norman, (from the Alfred Hitchcock movie) and it was a psychopath.

They tested Norman with inkblots, the old Rorschach test. When the “normal” AI saw a group of birds sitting on top of a tree branch Norman saw a man being electrocuted. When the “normal” AI saw a wedding cake Norman saw a man getting shot dead. Norman only saw horrifying images and captions so it sees something morbid in whatever image it looks at.

Microsoft had a similar experience with its Twitter bot “Tay” in 2016. It had to be shut down within hours of its launch when it started spewing hate speech, racial slurs and denying the holocaust. Note to the “parents” of AI constructs: Be aware of what your “children” are reading, watching and who they are hanging out with!

While it is improving at exponential rates AI is still in the early stages of being able to mimic the human brain. Just with the current capabilities AI is revolutionizing computing. In the long run I think that AI will have a net positive effect on humanity.

And remember always back it up!

Go back to the top