A few days ago, I read an article in Wired magazine about a computer system built by Google and partners that defeated a Grandmaster at Go, four games to one in a five-game series.* Go is a 2500 year old Chinese strategy game with exponentially more complexity and potential moves than Chess and, unlike Chess, has until now resisted the powers of computing competition and remained in the realm of the human mind. I read the this article with casual interest, but it stayed with me and as I continued to think about it, I began to shift to an examination of the short history of supercomputing and artificial intelligence (AI) and reflected on the implications of a technical revolution that has blown Moore's Law to bits.
In the 1970s and '80s, Cray Research (founded by Seymour Cray) was, in its heyday, building lightning fast machines used for high capacity scientific and commercial applications. In 1983, Danny Hillis founded Thinking Machines Corporation in order to build a massively parallel processing machine (known as the Connection Machine) that fundamentally altered the von Neumann architecture of computing. These pioneers, along with others, built systems that provided an enabling platform for the work spearheaded at MIT in the field of AI.
Artificial Intelligence research has been in place for 60 years now and, while remaining on the periphery of the consciousness of the average technology buff, it has produced remarkable, almost unbelievable, manifestations. The predictions of the AI thought leaders (Marvin Minsky, Herbert Simon, et. al.) have largely been realized through expert systems and other applications. The latest victory on a Go board signals a new breakthrough in the deep learning and application capabilities that today's computing power and neural networks allow. Right now, as unsettling as it is, , it is clearly possible to envision machines with greater rational intelligence than humans.
In surveying the inexorable rise of computing power, networking and the resulting AI progress, it is inevitable that people begin to think about the effect on society and our daily lives. Clearly, the world of human decision making (both quantitative and qualitative) is sure to be greatly altered. Extremists feel that whole industries will be transformed (including the military, education and the judicial system). Corporate management as we know it would be radically different with highly intelligent computers doing the work of scores of white collar workers. There are already robots in manufacturing facilities. The cost of a digital workforce would be much less than its current level. No healthcare, no pensions, no 401k to support. In fact, the world would enter the post-information age, where information and the ultra sophisticated decision support apparatus that results would be a given and the role of the worker would be in support of those systems and networks. It will be an economy based on the reach of computing.
However, thoughtful these views are, they leave out some of the most human of endeavors that give texture to the fabric of society. Here are three thoughts:
The arts, for example, will always be a human activity, even though the media used may (and will ) often be digital in nature. The expression of emotion and an examination of the human condition are what the arts are about and the beauty created by an artistic representation will always remain rooted in humanity.
In addition, while the role of computer -generated learning will sure increase (e.g. IBM's Watson), the role of teachers must also remain in the realm of human interaction. A teacher provides not only the explanation of the curriculum, but also counseling and empathetic support for the student's issues and learning strategies. In other words the role of educational mentoring reaches far beyond the delivery of information, but reaches into the psychological support of each developing student. That role cannot be performed even with the most powerful of AI systems.
Lastly, the democratic political process and the election of government leaders will (and should) remain in the control and with the consent of the body politic. Governing strategies will be greatly enhanced by the intervention of proliferated computing power, but leadership is a qualitative trait that engages and galvanizes people toward the gestalt of the nation. We follow those who lead and provide an emotional connection to the solutions provided by the expert systems at work. We elect our leaders - we don't calculate them.
The future of computing and artificial intelligence hold great promise for the future of society. While fundamentally altering the economic construct, it allows for a fundamentally new paradigm for the development and fulfillment of a meaningful career and provides a basis of knowledge that will guide, rather than supplant, the lives of human beings. I don't believe we will ever reach the stage of truly sentient computing and that being said, machines will never be able to replicate that most basic of human emotions - the capacity and need to give and receive love.
*CORRECTION: This post originally and stated incorrectly that Google's computer system won five games in a row. The human grandmaster, Lee Sedol, won the fourth game of the series.