Can machines think?
This is a question posed by Alan Turing in the 1950s that forever changed the science and history of Artificial Intelligence.
Now I’m pretty sure you’ve watched or heard at least a few interesting things about Artificial Intelligence (AI), as depicted in various science fiction films and series. While some of them tackles intriguing perspectives on human-like robots, the field of AI actually encompasses a lot more than that. From Google searches, computer programs, smart devices to virtual assistants such as Siri, and self-driving cars you see in tech documentaries, all these include the application of AI.
As you can see, the extensive range of AI makes this branch of computer science truly fascinating. If you understand its basic aspects and delve deeper to its compelling aspects, you will find that it has formed an important part of our lives.
That’s why in this reading, we will be discussing the various facets surrounding Artificial Intelligence that can give you a wider perspective on what it really is: through its history, general definition and scope, categories or subtypes, and applications or examples.
The history of artificial intelligence stretches way back to the paradigms of Philosophy concerning with the search for truth and knowledge, along with the ancient Greek myths of Antiquity. Like most sciences, the branch’ past records and chronicle of events are deeply-rooted in our history.
However, the more familiar concepts and perspectives we know of AI today encompasses various and continuous scientific developments that has spanned for decades.
For example, in the 1950s to 1970s, early research on AI explored issues on problem solving and symbolic methods that eventually led to the revolutionary work on computers mimicking basic human reasoning. Throughout its history, this has been its primary goal or essence, in creating artificial intelligence similar to that of humans’. Overall, the 1950s has sparked interests in thinking machines to which the essence of artificial intelligence will hinge on for generations.
In his published work, Computing Machinery and Intelligence, Alan Turing proposed the Turing Test as a measurement of machine intelligence. Within the same decade, Marvin Minsky and Dean Edmonds also built the first neural network computer. And soon enough, the phrase ‘artificial intelligence’ was officially coined by the father of AI, John McCarthy. Thereafter, machine learning became popular and was the focus of most AI research.
In the 2000s, more intelligent machines such as Siri, Alexa and Cortana are thoroughly developed in tech industries and are now extensively produced in the market. Today, in-depth research studies and breakthroughs on deep learning practically paves the way towards a promising future of AI for years to come.
As you can see, the field of AI started with a humble and notable beginning and soon bloomed into more specific areas of specialization. As the world progresses along technology and modernization, subfields under AI become even more sophisticated and we will definitely come across more groundbreaking research and innovations in the near future.
Throughout the years, the evolution of AI has constantly brought about significant changes and improvements to various industries such as technology, health care, education, business, and a lot more.
Artificial Intelligence is a branch under computer science that aims to understand ‘smart machines’, along with the mechanics involved in building or developing one.
AI has various subfields that branch out from more general subject matters to more specialized topics in catering to knowledge gaps on specific paradigms. Coupled with extensive debates and issues regarding the field, AI is an interdisciplinary science that utilizes multiple approaches to better understand its own subject of study. This leads to a non-singular definition that can accurately embody all, if not most, of its principles.
AI tries to answer Alan Turing’s question as mentioned earlier: can machines actually think? If they do, how exactly does that happen? How do they act or behave? And what exactly makes a machine intelligent?
The short answer is: yes, machines can think. We can constantly develop machines to think more and more like humans do.
Stuart Russel and Peter Norvig, in their book Artificial Intelligence: A Modern Approach published in 1995, states that “AI attempts to understand intelligent entities”. This is all the more reason for us to study and learn more about it. Unlike other branches of science that primarily deal with human intelligence, AI doesn’t stop at understanding these intelligent entities, but also involve with building them.
The authors also categorize AI systems into 4 categories, namely:
(1) systems that think like humans;
(2) systems that think rationally;
(3) systems that act like humans; and
(4) systems that act rationally.
In a broader prospect, AIs are smart systems that are programmed to have the capability of thinking, acting, behaving, and performing tasks intelligently and rationally without too much reliance and dependence on human instructions. Of course, they are created by humans, but as all these systems combine into an individual program, they can also think, act, and behave on its own.
If a device or an engine can manifest these thought processes, reasoning, and behavior, then you can safely call it a smart machine.
While these may sound abstract, these categories have been used as a framework that defined what AI is in journals and research studies ahead of Russel and Norvig’s groundwork. As a whole, this sets up a criterion in narrowing down the things scientists should study in the field and therefore help both professionals and laymen specify what an intelligent entity is.
If you look more closely around you, AI is everywhere. The applications of AI cover from a wide range of mundane and subtle uses to more ‘grandiose’ ones.
Some examples of AI are your smart virtual assistants such as Siri, Alexa, and Cortana. You may have also stumbled upon a few conversational bots online with virtual customer service representatives assisting you with their brand’s products and services.
AI is also applied to disease mapping and prediction tools such as the AI-driven tech made by Korea for effective contact tracing amidst the COVID-19 crisis. Under the health industry, personalized health care treatment recommendations made by AI are constantly optimized to effectively cater to every patient’s unique health needs. Even simple apps, such as those designing a diet and workout routine for you, are considered AI as they gather data to produce a more effective health training program.
Drone robots used for filming and surveillance are also improved and manufactured to become smarter and more advanced than previous models.
The algorithms in various internet sites you use are also arranged by smart AI to help maintain a personalized and familiar pattern, as seen in your social media feed or Google searches. Due to the prevalent of internet misinformation nowadays, AIs are also used to monitor and filter false and dangerous online information to prevent deceptions.
AI has 2 broad categories to which you can easily classify them: Weak AI and Strong AI.
Weak AI or Narrow AI performs in a more limited scope of human intelligence than Strong AIs do. Of course, they are still smart machines, but they are often programmed to operate specific tasks in a restricted or constrained structure. Examples include search algorithms, virtual assistants, and image recognition software.
Most Weak AI machines and programs are powered by a set of carefully designed algorithms such as machine learning and deep learning.
To illustrate a simple difference between the two, machine learning puts in data and statistical techniques into a machine to help it ‘learn’ and gets better at a specific task, thus creating its own program. Deep learning, on the other hand, is a subset of machine learning that utilizes biologically-inspired neural network structures to program machines. Because of its structure, machines have the capability to go ‘deeper’ in its learning and creates its own neural data connections within the program.
Other types of narrow AI include: neural networks (that utilizes similar mechanics behind deep learning), natural language processing, computer vision, and cognitive computing.
Strong AI or Artificial General Intelligence (AGI), aims higher in creating more intensive and extensive human-level intelligence. However, a lot of research on strong AIs or AGIs raises ethical implications and concerns. This type of AI is often portrayed in a misleading way through series and films giving narratives about super smart robots taking over humans and the world.
While the topic continues to spark debates, the future of Strong AIs is quite promising. As we learn more about intelligent machines, we’ll also learn more about ourselves and humanity that can hopefully provide us with a wider perspective on what it means to be an intelligent entity.