Artificial Intelligence — An Introduction
Artificial Intelligence is a concept simultaneous fantasized by storytellers, debated by philosophers and investigated by scientists. Formally, Artificial Intelligence (AI) is a branch of computer science that deals with the study of intelligence machines. The term artificial intelligence was first coined by John McCarthy in his 1955 proposal for the 1956 Dartmouth Conference, the first academic conference on AI. McCarthy believed that “every aspect of learning or any other feature of intelligence can in principle be so precisely described that a machine can be made to simulate it.”
AI remains one of the most elusive subjects in computer science, in part due to the ambiguous nature of intelligence and the philosophical debates behind a machine’s ability to “think”. Undoubtedly a computer is capable of logic, but can a machine “think”? John Searle, a philosopher, presented a thought experiment called the Chinese Room in a paper published in 1980. The experiment stated that a person is passed questions written in Chinese from another room. The person does not understand Chinese, but has rules around which questions correspond to which answers, and what are the rules of the answers. In the end, the person could take the Chinese questions, and using a series of rules, write a Chinese answer to pass to the outside of the room. Searle argued that while the respondent was able to answer questions in Chinese, they did not understand their own responses. Rather, they were simulating the ability to understand Chinese, which is what a machine does using programs and codes.
Therefore, if artificial intelligence is a series of rules and logic, can a machine “think” without “understanding”?
In 1950, English mathematician Alan Turing developed what we now refer to as “The Turing Test”. In a common interpretation of the The Turing Test, an interrogator chats with two subjects over text messages. One subject is a computer and the other subject is a human. A computer has passed the Turing Test if the interrogator cannot distinguish between the human and computer. The test, or the “Imitation Game” as it was called by Turing, was proposed as a simple test to prove that machines could think. For the more visual readers, The Imitation Game is a brilliant film depicting the life of Alan Turing. In 2014, Eugene Goostman, a computer program simulating a 13-year-old boy, convinced 33% of human judges at the Royal Society in London that he was human. The official threshold for passing the Turing Test was to convince 30% of human judges, making Eugene the first computer program to pass the Turing Test. Does this mean we should start welcoming our robot overlords?
While Eugene passed the Turing Test, he was programmed to be a 13-year-old boy from Ukraine and thus had limits in knowledge and language that could be explained by his young age and having English as his second language. Today, we are surrounded by examples of AI with varying levels of sophistication. Basic artificial intelligence uses code that detects certain cues, then comes up with a reaction. For example, your email inbox likely automatically filters out spam. Basic spam filters register certain words or cues in an email to separate it into spam or real email. Basic artificial intelligence uses code that detects certain cues, then comes up with a reaction. A classic “If-then”, or conditional statement.
More sophisticated email inboxes take it to the next level by using machine learning algorithms to personalize your spam filter. While AI is considered the broader concept of machines being able to “think”, Machine Learning is an application of AI that allows programs to learn for themselves. Since everyone’s definition of spam is different, and since spammers continuously change their messages, spam filters must learn what is spam from different cues. These machine learning algorithms continuously collect data to refine the definition and cues of spam. Every time you click “Report Spam” or “Not Spam”, you’re teaching the program and refining the experience. As a result, less than 0.1% of emails in your average Gmail folder is spam and less than 0.05% of emails in your spam folder are wanted emails.
Today, the latest development in AI involves Deep Learning. It can be considered a subfield of machine learning and involves feeding a computer system a lot of data to be processed into an output. The genius of deep learning comes from the data being processed through hierarchical levels of neural networks, that were designed to mimic the biological nervous system. For a more detailed explanation, check out this video!
Deep learning has refined and improved multiple applications, including image recognition, speech recognition and generation, and self-driving car.
With AI advancing at breakneck speeds, should we start fearing a dystopian future where machines turn against us? If you are a participating member of modern society, you are already surrounded by many examples of artificial intelligence, such as Siri (Apples personal assistant you can summon from your Apple product with your voice), Alexa (Google’s personal assistant that tailors its service to you the more you use Google products) and Amazon (with it’s “people that bought this also bought this feature”). Credit card companies use AI to detect potentially fraudulent activities. There is no reason to fear the future of AI considering it has already been around for years and you are probably already using some form of it.
For those who are still not convinced, countries are collaborating to discuss policies for safeguards around AI, ethics and privacy. What development of Artificial Intelligence are you most excited about?
On February 21, the topic of discussion will be the exploration of Artificial Intelligence and Machine Learning; how can this new technology benefit businesses in the future?
Tickets are currently on sale; for more information visit ChicGeekYYC or Universe for ticket registration.
