By Stephen Eick
Take a brief moment and picture the world prior to the Industrial Revolution. Few people yearn for times without our modern amenities, our technology. We safely conclude that the world ultimately benefits from the development of innovative tools, especially electronic tools. However, this exact moment in history holds the most rapid development humanity has ever experienced. The technological torrent of our time drowns the important conversations and observations needed as we enter an age where humanity is augmented with robotics and artificial intelligence (AI).
To begin, what is robotics? “[It] is an extension of the Industrial Revolution,” says Michael Zinn, a Professor of Mechanical Engineering here at the University of Wisconsin-Madison. Professor Zinn researches minimally-invasive medical robotics, discovering superior ways to treat patients through applications of robotic systems. It is natural for humans to make a task easier through the development of tools. A robot is just a tool to accomplish societal goals and make the human existence better. It is controlled through of a series of controlled electrical impulses which translate mechanical systems with little-to-no external intervention. However, many robots do not possess intelligence; the exact sequence of electrical impulses are predetermined by a human trained to create the desired motion. Based on this information, we can conclude, say, the Bellagio fountain in Las Vegas is a robotic system but, like most other robotic systems, does not possess any sort of intelligence.
What happens when robots are able to make decisions, though? When an electromechanical system possesses the ability to perceive its environment and respond accordingly, one can say the system possesses a certain level of “intelligence.” However, we must ask if the system is following a predetermined list of commands to escape the situation rather than dynamically solving the problem. The average assembly-line robotic system is programmed to silently perform a single task as long as its motors move. We see a different scenario, however, when a robotic system can be taught a task through human-to-robot training sessions. A well-known robot of this sort is known as Baxter, a robot with a face and two arms that was introduced in 2012 by the company Rethink Robotics. Unlike standard robots, Baxter can learn new tasks. After a human guides Baxter’s arms through the motions of the task, it will be able to perform the task. Most impressively, Baxter can tolerate slop and mistakes in the process such as items not placed in the right location and dropping a tool it’s holding. In this sense, Baxter can learn and adapt to its environment. Compared to the standard robotic system, we see both static, rigid intelligences and dynamic, malleable intelligences.
Or so we say. Can this “dynamic” intelligence actually learn anything? To answer using our example, no, Baxter can’t learn dynamically. Baxter is designed for material handling in industrial environments — to humans, a low-skill job — and subsequently would not need to possess the cognition necessary for, say, cooking; a superbly difficult task. Baxter possesses what is known as constrained artificial intelligence, in that it can only solve a certain class or subset of problems. These intelligent systems are finding broader applications in the form of autonomous (self-driving) cars. Compared to Baxter, an autonomous car has intelligence that is much less constrained (the car must navigate a wider set of challenges), but still has limitations (the car still cannot cook a meal). When an artificially-intelligent system can operate at par with humans, the system can be said to possess general artificial intelligence. A general artificial intelligence is entirely feasible; even back in 1956, pioneering AI researchers at a conference at Dartmouth College proclaimed that “every aspect of learning or any other feature of intelligence can be so precisely described that a machine can be made to simulate it.” Another way of putting that statement is: if a human can learn it, so can a machine. While a truly general artificial intelligence does not yet exist, the puzzle pieces which will eventually complete that enormous puzzle are being built right now.
Even systems of constrained intelligence such as Baxter and autonomous cars will leave an impression on the future. In fact, the autonomous car may well be looked at as a pivotal moment in the history of humanity. “I’m just amazed at how fast [the autonomous car] has gone from something that seemed impossible to something that is practically here already,” says Professor Zinn. Even for a person who researches robotics for a living, the development of the autonomous vehicle comes off as incredible, exploding within the last fifteen years. According to members of the Institute of Electrical and Electronics Engineers (IEEE), autonomous cars aren’t expected to capture 75% of the vehicle market until 2040. However, their imminent arrival could lead to the elimination of many of the taxi and semi-truck drivers in the United States.
But is the elimination of driving and, more broadly, low-skill jobs really a bad thing? For sure, the nostalgia surrounding manual driving will remain strong, and some will remain adamantly opposed to relinquishing the exhilaration of the open roads. “I think the biggest concern with robotics is them taking jobs away,” Professor Zinn says. “And they do. From a macroeconomic point of view they don’t, but locally they certainly do.” As society has advanced and more opportunities arose, the occupation of semi-truck driver has experienced a severe decline in applicants. This shortage of drivers, coupled with the highest rate of work-related deaths (according to a Bureau of Labor Statistics report, 835 drivers experienced fatal injuries in 2014) will force the trucking industry to increase the adoption of autonomous semi-trucks whether they like it or not. Definite economic and societal concerns exist when companies replace workers with robotic systems, but the elimination of low-skill jobs could force society to pursue higher opportunities. While robotics may eliminate jobs, they are jobs society is rapidly losing interest in holding.
Technological progress may be disruptive, but it is progress nonetheless. Up until now, our robots augmented the assembly line, replacing workers with machines that perform repetitive tasks better than humans do. Robots will gain intelligence and self-governance, allowing them to make key decisions without our input. We humans, always vigilant and self-preserving, must ensure their applications strengthen, rather than damage, humanity. The same follows for artificial intelligence; as the current limited patchwork of AI coalesces into a powerful entity of general intelligence and self-governance, we must take care to guard that which is distinctly human while embracing the immense capabilities wrought by artificial intelligence.