Can we create true artificial intelligence?
Artificial intelligence is right here with us. In research centers and laboratories around the globe, smart AIs are springing to life. They are taking over literally every aspect of our lives from beating the world’s best human mind in a chess game to learning how to drive millions of cars in our cities. Some are currently being deployed in our hospitals and emergency rooms where they’re saving lives even more than most physicians would ever dream.
Others get integrated into smart homes where they regulate the temperature of your homes, ensure the bath water is just right for you, and even play your favorite music when you wake up in the morning. Countless numbers of these “little” intelligence are being programmed; they would continue to get smarter and become more prevalent. They would even get better than us, but they would never be us.
Wait. Isn’t that what artificial intelligence is all about?
To develop intelligence that is capable of mimicking every aspect of the human being – the ultimate goal of evolving into a self-aware, empathic, self-learning entity.
So why couldn’t they be like us then?
You see, the human brain is such a complicated mass of neurons that even with all the latest advancement in technology, we haven’t fully understood how it works. Hence, it’s going to be unlikely that anyone would develop an AI that’s fully able to stimulate the human brain at least not anytime soon. However, in spite of my not so enthusiastic outlook about our readiness and capability to develop true artificial intelligence, there are some who believed if we can get a few hurdles out of the way, we would undoubtedly create true AIs in a couple of decades.
One such challenge is getting the computer to pass the Turing test. Back in 1950, after Alan Turing delivered his seminal paper on building a “thinking” computer, artificial intelligence scientists have made it their holy grail to build machines that are as closely related to humans as possible. To pass the test, the computer must pass off as human when communicating with a real human.
It sounds straightforward, isn’t it?
But the reality is creating a computer that is powerful enough to stimulate the human’s natural language involves developing a complex set of an algorithm that replicates the brain’s neuron patterns; develop a language engine that essentially synthesizes natural language, and of course, generate sufficient energy to run this machine.
Now, this doesn’t seem so easy anymore, right?
Other challenge scientists must overcome in their quest for the true artificial intelligence is to match human’s ability to interpret visual information. You see, computers are currently miles behind in their capability to process images when compared to the human brain.
An excellent example of this in play is the spam check on most websites that require humans to read and input a series of random letters and numbers before accessing the site. That’s a reverse Turing test where the computer is trying to differentiate between real human users and bots. It’s easier for a human to understand the warped letters while a computer finds it extremely unreadable.
This demonstrates just how poor current computer visual processing power is. As it is computers read images pixel by pixel and lacks the ability to bring them all together into a seamless well-integrated data.
So back to the question, can we create true artificial intelligence?
The answer is yes with a caveat though; it’s not going to happen, say in the next 20 years or even 30 years.
Of course, I understand the giant strides made so far; however, the reality is we haven’t figured out what the essence of intelligence is, and until we do – creating a human-level artificial intelligence will still be elusive.