AI research funding was a roller-coaster ride from the mid-1960s through about the mid-1990s, experiencing incredible highs and lows. By the late 1990s through the early part of the twenty-first century, however, AI research began a resurgence, finding new applications in logistics, data mining, medical diagnosis, and numerous areas throughout the technology industry. Several factors led to this success.

Computer hardware computational power was now getting closer to that of a human brain (i.e., in the best case about 10 to 20 percent of a human brain).

• Engineers placed emphasis on solving specific problems that did not require AI to be as flexible as a human brain.
• New ties between AI and other fields working on similar problems were forged. AI was definitely on the upswing. AI itself, however, was not being spotlighted. It was now cloaked behind the application, and a new phrase found its way into our vocabulary: the “smart (fill in the blank)”—for example the “smartphone.” Here are some of the more visible accomplishments of AI over the last fifteen years.
  • In 1997 IBM’s chess-playing computer Deep Blue became the first computer to beat world-class chess champion Garry Kasparov. In a six-game match, Deep Blue prevailed by two wins to one, with three draws. Until this point no computer had been able to beat a chess grand master. This win garnered headlines worldwide and was a milestone that embedded the reality of AI into the consciousness of the average person.
  • In 2005 a robot conceived and developed at Stanford University was able to drive autonomously for 131 miles along an unrehearsed desert trail, winning the DARPA Grand Challenge (the government’s Defense Advanced Research Projects Agency prize for a driverless vehicle).
  • In 2007 Boss, Carnegie Mellon University’s self-driving SUV, made history by swiftly and safely driving fifty-five miles in an urban setting while sharing the road with human drivers and won the DARPA Urban Challenge.
  • In 2010 Microsoft launched the Kinect motion sensor, which provides a 3-D body-motion interface for Xbox 360 games and Windows PCs. According to Guinness World Records since 2000, the Kinect holds the record for the “fastest-selling consumer electronics device” after selling eight million units in its first sixty days (in the early part of 2011). By January 2012 twenty-four million Kinect sensors had been shipped.
  • In 2011, on an exhibition match on the popular TV quiz show Jeopardy!, an IBM computer named Watson defeated Jeopardy!’s greatest champions, Brad Rutter and Ken Jennings.
  • In 2010 and 2011, Apple made Siri voice-recognition software available in the Apple app store for various applications, such as integrating it with Google Maps. In the latter part of 2011, Apple integrated Siri into the iPhone 4S and removed the Siri application from its app store.
  • In 2012 “scientists at Universidad Carlos III in Madrid…presented a new technique based on artificial intelligence that can automatically create plans, allowing problems to be solved with much greater speed than current methods provide when resources are limited. This method can be applied in sectors such as logistics, autonomous control of robots, fire extinguishing and online learning” (www.phys.org, “A New Artificial Intelligence Technique to Speed the Planning of Tasks When Resources Are Limited”).

The above list shows just some of the highlights. AI is now all around us—in our phones, computers, cars, microwave ovens, and almost any consumer or commercial electronic systems labeled “smart.” Funding is no longer solely controlled by governments but is now being underpinned by numerous consumer and commercial applications.

The road to being an “expert system” or a “smart (anything)” focused on specific well-defined applications. By the first decade of the twenty-first century, expert systems had become commonplace. It became normal to talk to a computer when ordering a pharmaceutical prescription and to expect your smartphone/automobile navigation system to give you turn-by-turn directions to the pharmacy. AI clearly was becoming an indispensable element of society in highly developed countries. One ingredient, however, continued to be missing. That ingredient was human affects (i.e., the feeling and expression of human emotions). If you called the pharmacy for a prescription, the AI program did not show any empathy. If you talked with a real person at the pharmacy, he or she likely would express empathy, perhaps saying something such as, “I’m sorry you’re not feeling well. We’ll get this prescription filled right away.” If you missed a turn on your way to the pharmacy while getting turn-by-turn directions from your smartphone, it did not get upset or scold you. It simply either told you to make a U-turn or calculated a new route for you.

While it became possible to program some rudimentary elements to emulate human emotions, the computer did not genuinely feel them. For example the computer program might request, “Please wait while we check to see if we have that prescription in stock,” and after some time say, “Thank you for waiting.” However, this was just rudimentary programming to mimic politeness and gratitude. The computer itself felt no emotion.

By the end of the first decade of the twenty-first century, AI slowly had worked its way into numerous elements of modern society. AI cloaked itself in expert systems, which became commonplace. Along with advances in software and hardware, our expectations continued to grow. Waiting thirty seconds for a computer program to do something seemed like an eternity. Getting the wrong directions from a smartphone rarely occurred. Indeed, with the advent of GPS (Global Positioning System, a space-based satellite navigation system), your smartphone gave you directions as well as the exact position of your vehicle and estimated how long it would take for you to arrive at your destination.

Those of us, like me, who worked in the semiconductor industry knew this outcome—the advances in computer hardware and the emergence of expert systems—was inevitable. Even consumers had a sense of the exponential progress occurring in computer technology. Many consumers complained that their new top-of-the-line computer soon would be a generation behind in as little as two years, meaning that the next generation of faster, more capable computers was available and typically selling at a lower price than their original computers.

This point became painfully evident to those of us in the semiconductor industry. For example, in the early 1990s, semiconductor companies bought their circuit designers workstations (i.e., computer systems that emulate the decision-making ability of a human-integrated circuit-design engineer), and they cost roughly $100,000 per workstation. In about two years, you could buy the same level of computing capability in the consumer market for a relatively small fraction of the cost. We knew this would happen because integrated circuits had been relentlessly following Moore’s law since their inception. What is Moore’s law? I’ll discuss this in the next post.

Source: The Artificial Intelligence Revolution (2014), Louis A. Del Monte

Image: iStockPhoto.com (licensed)

While the phrase “artificial intelligence” is only about half a century old, the concept of intelligent thinking machines and artificial beings dates back to ancient times. For example the Greek myth “Talos of Crete” tells of a giant bronze man who protected Europa in Crete from pirates and invaders by circling the island’s shores three times daily. Ancient Egyptians and Greeks worshiped animated cult images and humanoid automatons. By the nineteenth and twentieth centuries, intelligent artificial beings became common in fiction. Perhaps the best-known work of fiction depicting this is Mary Shelley’s Frankenstein, first published anonymously in London in 1818 (Mary Shelley’s name appeared on the second edition, published in France in 1823). In addition the stories of these “intelligent beings” often spoke to the same hopes and concerns we currently face regarding artificial intelligence.

Logical reasoning, sometimes referred to as “mechanical reasoning,” also has ancient roots, at least dating back to classical Greek philosophers and mathematicians such as Pythagoras and Heraclitus. The concept that mathematical problems are solvable by following a rigorous logical path of reasoning eventually led to computer programming. Mathematicians such as British mathematician, logician, cryptanalyst, and computer scientist Alan Turing (1912–1954) suggested that a machine could simulate any mathematical deduction by using “0” and “1” sequences (binary code).

The Birth of Artificial Intelligence

Discoveries in neurology, information theory, and cybernetics inspired a small group of researchers—including John McCarthy, Marvin Minsky, Allen Newell, and Herbert Simon—to begin to consider the possibility of building an electronic brain. In 1956 these researchers founded the field of artificial intelligence at a conference held at Dartmouth College. Their work—and the work of their students—soon amazed the world, as their computer programs taught computers to solve algebraic word problems, provide logical theorems, and even speak English.

AI research soon caught the eye of the US Department of Defense (DOD), and by the mid-1960s, the DOD was heavily funding AI research. Along with this funding came a new level of optimism. At that time Dartmouth’s Herbert Simon predicted, “Machines will be capable, within twenty years, of doing any work a man can do,” and Minsky not only agreed but also added that “within a generation…the problem of creating ‘artificial intelligence’ will substantially be solved.”

Obviously both had underestimated the level of hardware and software required for replicating the intelligence of a human brain. By setting extremely high expectations, however, they invited scrutiny. With the passing years, it became obvious that the reality of artificial intelligence fell short of their predictions. In 1974 funding for AI research began to dry up, both in the United States and Britain, which led to a period called the “AI winter.”

In the early 1980s, AI research began to resurface with the success of expert systems, computer systems that emulate the decision-making ability of a human expert. This meant the computer software was programmed to “think” like an expert in a specific field rather than follow the more general procedure of a software developer, which is the case in conventional programming. By 1985 the funding faucet for AI research was reinitiated and soon flowing at more than a billion dollars per year.

However, the faucet again began to run dry by 1987, starting with the failure of the Lisp machine market that same year. The Lisp machine was developed in 1973 by MIT AI lab programmers Richard Greenblatt and Thomas Knight, who formed the company Lisp Machines Inc. This machine was the first commercial, single-user, high-end microcomputer and used Lisp programming (a specific high-level programming language). In a sense it was the first commercial, single-user workstation (i.e., an extremely advanced computer) designed for technical and scientific applications.

Although Lisp machines pioneered many commonplace technologies, including laser printing, windowing systems, computer mice, and high-resolution bit-mapped graphics, to name a few, the market reception for these machines was dismal, with only about seven thousand units sold by 1988, at a price of about $70,000 per machine. In addition Lisp Machines Inc. suffered from severe internal politics regarding how to improve its market position, which caused divisions in the company. To make matters worse, cheaper desktop PCs soon were able to run Lisp programs even faster than Lisp machines. Most companies that produced Lisp machines went out of business by 1990, which led to a second and longer-lasting AI winter.

In the second segment of this post we will discuss: Hardware Plus Software Synergy

Source: The Artificial Intelligence Revolution (2014), Louis A. Del Monte