Artificial Intelligence, or AI, has been popularized by science fiction for decades. Although it’s entertaining to imagine hordes of killer robots, the truth about it is far more down to earth. To understand why robots will probably not enslave us all in the next few decades, it’s important to know where AI came from, and where it’s going.
The very first computer was built in the 1940’s. It had less computational power than a calculator and was roughly the size of a small house. In 1956, a workshop held at Dartmouth College explored the possibility of having these new computers programmed with intelligence (Datrmouth University, 2006). Over the next half century, excitement and funding for artificial intelligence peaked and dipped. As soon as there was a new breakthrough, rivers of money would flow into advancing the field, usually with comparatively few real results. This was due to the size and limited processing power of computers at the time.
What Is Moore's Law?
The law suggests that the number of transistors per square centimeter on a microchip would double at least every 18 months. Invetopedia states that Moore's law predicts that this trend will continue into the foreseeable future. Although the pace has slowed, the number of transistors per square inch has since doubled approximately every 18 months. As of this year, it is possible that Moore’s Law will be disproven. That means that we are reaching the physical limit of how many transistors can be attached to a microchip. As one could imagine, this also means that modern microchips are very, very powerful.
This steady increase in microchip power has led to the current state of AI. In the first years of the 21st century, when machine learning algorithms began to show improved productivity in the industry, a new wave of interest hit.
While the innovations and discoveries that have been made are incredible, we are still far from true intelligence. Current AI deployments are usually very specialized. For example, the new RankBrain AI that Google uses to determine the usefulness of a website can process billions of events each minute, but it can’t paint a picture. It is a genius in a single aspect, but still far from truly intelligent. Other applications for AI are advanced mathematics, physics modeling (MELVIN Choi, 2016), statistical analysis and probability calculations.
For an AI to be considered truly intelligent, it would need to be self-aware and have the ability to be self-determining. Since we don’t really understand much about either of these phenomena in ourselves, the likelihood that we could replicate it in a computer is fairly small. That being said, AI programs can do a fantastic job of simulating intelligence, or “acting” in an intelligent manner. This was proven when the Google Deepmind AI, Alpha Go, defeated the current world champion at the game “Go” (Byford, 2016). Many years ago, an AI defeated a chess grandmaster. That was done by probability analysis. The number of potential moves in the game “Go” is astronomical, probability analysis could never have worked. Instead, Alpha Go simulated human intuition.
While the ability to simulate intuition is a giant leap forward in AI research, it’s still not true intelligence in the natural sense. Natural intelligence is a concept that the human race has been struggling with for the past few thousand years. We consider it to be the ability to be aware of ourselves in relation to our environment. It is consciousness. We are still far from developing machines that are able to consider their own existence and make decisions based on abstract concepts like fear, love or hate.
On the other hand, it’s possible that we could create a new form of intelligence altogether. Advances in the field of quantum computing could one day make the processing power of our most advanced processors seem as limited as the first computer ever built. If we were able to create an AI that self-improved, coupled with the power of a quantum processor, it’s possible that it would evolve into something we could never have imagined.
The field of artificial intelligence is exciting and terrifying. For us to understand and guide it, education in the fields of mathematics, machine learning and computer science need to become the main focus of higher education systems.
Byford, S (2016). Google's DeepMind defeats legendary Go player Lee Se-dol in historic victory. Retreived from https://www.theverge.com/2016/3/9/11184362/google-alphago-go-deepmind-result
Scientific American (2016). Physicists Unleash AI to Devise Unthinkable Experiments
Dartmouth University (2006). The Dartmouth Artificial Intelligence Conference: The Next Fifty Years
Retrieved from https://www.dartmouth.edu/~ai50/homepage.html