Last February, viewers across North America watched as the unthinkable unfolded on the decades-old quiz show, Jeopardy!: A computer named Watson handily beat two of the best human competitors to have ever appeared on the program.
What they might not have realized is they were also witnessing a major development in a branch of computer science that is changing the way we work, communicate, shop, bank and live.
Artificial intelligence, a broad term loosely defined as the ability of computers to behave like humans, has experienced a major surge in recent years as scientists inch ever closer toward the breakthroughs they hope will allow machines to eventually learn and adapt like humans.
The surge is due largely to the fact AI research has become increasingly focused in recent years on developing practical applications that can be used by businesses or governments, which has generated considerable excitement and an infusion of public and private sector cash.
AI combines the fields of computer sciences, mathematics, biology, ethics, psychology and even philosophy. Scientists in labs around the world are working to develop machines capable of logic and reasoning, analysis and intuition.
Although many consumers may not realize it, they likely come into contact with aspects of AI daily, from Google results that appear as users type search requests to book recommendations made based on an individual’s buying habits on Amazon.
Many people associate AI with the plot of sci-fi movies: Humans are at war with self-aware, intelligent computer systems, such as those featured in the Terminator and The Matrix franchises. In reality, AI remains in many respects a long way from developing machines that can meet or surpass the capabilities of the human mind.
But important developments and growing research investments have many in the field convinced that day will come, presenting enormous opportunities and the need for complex privacy and security controls.
Breakthroughs in AI could allow for countless developments with the potential to reshape the world, such as robots that can replace human soldiers, search engines that can understand the meaning of text to generate more accurate results and advertising campaigns tailored to a consumer’s individual tastes.
“Basically, anything people can do in the long run, machines are going to do better and cheaper,” said Geoffrey Hinton, a computer science professor at the University of Toronto who is considered one of the world’s leading experts on artificial intelligence. “The question is just how long that run is.”
The increasing “humanization” of computers also prompts troubling questions about privacy, security and the rise of a machine that can “outsmart” humans. For instance, many consumer advocates are already concerned about the amount of personal information computer programs are able to collect and store, something that would likely only increase under the expanded data-mining capabilities artificial intelligence could offer.
Others even cast doubt on the future of artificial intelligence and the ability of scientists to overcome the field’s biggest challenges. During a recent panel discussion hosted by the Massachusetts Institute of Technology, some of the leading minds in the world of AI argued the field has become too fragmented and narrowly focused to achieve the ultimate goal of creating truly intelligent machines.
The field of AI has become increasingly large and complex in recent years, leading to the creation of numerous sub-specialties and narrow areas of study. Now, instead of trying to answer fundamental questions about how to make machines more intelligent, many computer scientists are using algorithms to solve highly specific problems.
Despite the concerns, developments in artificial intelligence are generating considerable excitement in business circles because of the seemingly limitless possibilities it has enabling greater efficiency at much lower costs.
Reports surfaced earlier this month suggesting Apple Inc. will unveil new voice recognition capabilities, an important area of artificial intelligence, on the next version of the company’s iOS. Google is lobbying to make Nevada the first state to legalize driverless cars on public roads, as it has developed just such a car. And iRobot Corp., a robotic technology company, recently announced it had received a $230-million (U.S.) contract from the U.S. Navy.
Investment and interest in AI is ramping up largely because of the growing dominance of a sub-field called machine learning. It uses complex algorithms that allow computers to look for patterns and make decisions when exposed to large amounts of data. The more data the computer is exposed to, the more accurate its abilities become.
For example, Google’s computer system “learns” how to improve Web page rankings based on which results large numbers of users click on. And Apple, Amazon and Netflix use a variant of these algorithms to recommend future purchases.
“These are very core technologies for their business,” said Yoshua Bengio, professor in the department of computer science and operations research and Canada Research Chair in statistical learning algorithms at the University of Montreal. “It’s not perfect, there’s a lot more to be done, but it works much better than what we had before and it’s also more similar to how we think humans get to be intelligent in the first place.”
Two of the major challenges facing AI researchers are getting computers to have human-like vision and be able to understand natural language, tasks many people take for granted, but are extremely difficult to explain in a computer program.
Which explains why so much excitement was generated when Watson, a computer created by IBM, was able to answer questions posed by a human on Jeopardy! Many experts say it’s the biggest development in AI since an IBM computer named Deep Blue beat a world champion in chess.
“To understand questions to answer is actually very hard,” Prof. Bengio said. “I’m really confident there will be a lot of progress in that direction in the next few years.”
