Artificial Intelligence

Artificial Intelligence

What Is Artificial Intelligence?

There’s so much talk these days about artificial intelligence, but most people tend to focus on the practical uses, rather than the technology behind it.

When you hear about AI, you think about the various robots you’ve seen in the movies, or the robot housekeeper that one day will cook dinner for you and do the dishes. That’s all very good, but let’s find out what powers these robots—what do they use for a brain?

How Do You Define Artificial Intelligence?

artificial-intelligence-cloudOne of the best-known definitions of artificial intelligence describes AI as “the science and engineering of making intelligent machines [and] intelligent computer programs”.

The definition was put forth by John McCarthy in a 2004 paper of seminal importance. McCarthy goes on to address a crucial problem—how do you define machine intelligence? Does it have to mimic human intelligence or can machines be allowed or encouraged to develop their own way of thinking?

This debate goes back to 1950 when the great Alan Turing put the question bluntly: Can machines think? He also devised the famous Turing Test which uses a series of questions and answers meant to an interviewer to distinguish between a machine and a human being.

The implication was that a machine cannot develop human intelligence. In his 2004 paper, McCarthy has an interesting point of view: “AI does not have to confine itself to methods that are biologically observable”.

In other words, machines can and most probably will develop ways of thinking that escape us. They will get a good grasp of human thought patterns but also use their own intelligence.

For guys like Turing, AI would define systems that act and think like humans, but modern science is looking to create systems or software that act and think rationally, even if it’s not rational by human standards.

Weak AI vs Strong AI

Some people are smarter than others and the same could be said about machines. It all depends on the type of software that powers them.

Weak Artificial Intelligence

The term refers to systems built to perform a specific task. The machine focuses on a narrow pre-programmed task and that’s all. You may be surprised to learn that, although they are called “weak”, these types of AI are what powers systems like Amazon’s Alexa, Siri, or autonomous vehicles.

Strong Artificial Intelligence

artificial-intelligence-strong-aiIf a system smart enough to drive an autonomous car is weak, what is a strong AI capable of? Strong AI refers to Artificial General Intelligence (AGI) and Artificial Super Intelligence (ASI).

For the time being, both are mostly theoretical. We’re not there yet, but we may get there sooner than we’d expect.

Artificial General Intelligence refers to a machine capable of understanding the human thought process and imitating it. What’s more interesting is that an AGI machine would be self-aware and have the ability to solve problems unassisted, learn on its own, and plan for the future.

Artificial Super Intelligence is about Hal, the computer from 2001: A Space Odyssey. It doesn’t have to be a computer gone rogue, but that’s an outcome you cannot exclude.

An ASI machine would be smarter than humans, at least as far as its technical and reasoning functions are concerned. As such machines would also have a high degree of self-consciousness, they might one day ask themselves why are they taking orders from “dumb humans”.

Deep Learning vs. Machine Learning

Artificial Intelligence systems are defined by their capacity to learn, first by using algorithms created by humans, but then improving on them as they get smarter.

Intelligent systems use simple machine learning or deep learning. Deep learning uses neural networks that are linked just like the neurons in a human’s nervous system.

Deep neural networks have an input layer that receives data, several hidden layers that process said data, and an output layer. They don’t need any human assistance as the system automates most of its operations.

It’s the output layer in a node that connects to the input layer in the next node and transmits information.

Additional Artificial Intelligence Resources

Back to top button