The Evolution of Artificial Intelligence

By Matt Zhang

From Talos, the mythical automaton, to Frankenstein, in Mary Shelley’s book of the same name, artificial intelligence has been one of humanity’s utopian dreams throughout history. As we progress further towards the uncertainties of the future, and as our technological advances increase at an exponential rate, the possibility of truly self-thinking machines treads closer and closer to reality. How should we prepare ourselves? What may be the consequences? These are all questions that we must ask ourselves.

Historically, artificial intelligence has generally been used to theorize about machines and computational thinking, mostly revolving around answering questions or used as the basis for a few myths, as depicted in Greek mythology. However, these theories all relied on the assumption that human thought could be represented and computed, what we now know as computers, through the binary computational system. Eventually, as technology improved, scientists attempted to achieve more lofty goals, such as the creation of an artificially intelligent brain. Slowly, the research into AI continued, with the creation of the Turing test, a measure of the computer’s capacity to “think”, as well as the invention of programs to play complex board games such as chess. After a rather stagnant period in the late 1900s, AI saw a resurgence in the 21st century. Machines such as Deep Blue were created, with an estimated calculating capacity of 2 million moves a second. Self-driving cars and machine learning programs have also been created in the new century, introducing staggering amounts of innovation with them. As we move further into the modern age, the possibilities really are endless.

AI can be classified into several categories, artificial narrow intelligence, artificial general intelligence, and artificial superintelligence. These three refer to the strength of artificial intelligence, and its capabilities in terms of processing, calculation and thought. Artificial narrow intelligence refers to the usage of AI in specific tasks. Generally speaking, this type of AI is quite weak. The second type, general AI, refers to fairly strong AI, being able to solve many problems. This type of AI is what is referred to as “conscious”, the type typically depicted in popular media. It is autonomous and can think and devise problems and solutions for itself, just like a human being would. The final type of AI is called superintelligent AI, being capable of innovation and much more. In terms of scientific innovation, superintelligent AI would far exceed any possible human, likely being able to create more advanced AI than itself. A common theory is that these AIs would be able to become more and more intelligent until their functionalities become unfathomable to humans. 

However, there is not much need for concern. Currently, humanity is still predicted to be a few decades away from general artificial intelligence, meaning consciousness is still a while away. As of now, we can still reap all the benefits without caring about any of the consequences. Yet the day will come, and we will then need to rethink how we view artificial intelligence. Will we allow it to flourish, and eventually lead to superintelligence? The gap is quite small, given AI’s predicted constructive capabilities, and once the door is opened, it is impossible to turn back. Even more terrifying are the potential consequences. On one hand, the amount of technological innovation that can be brought with AI is essentially unfathomable. The consequences are of an equal scale. If we fail to regulate AI correctly, the effects could very well be what is seen in science fiction movies such as The Terminator or the Matrix. Even more concerning is the use of such technology, especially if it falls into the wrong hands. Given the tendency of human nature to lean towards innovation, as soon as the barrier is reached, it will likely be broken.  As of now, we can rest early, but for how long? Will we be ready? Only time will tell.