AI: A Timely Reminder

Over the past few years people—and governments—the world over have become increasingly enamored of AI. While some view AI as an economic game changer, the catalytic agent for a coming golden economic age, others are more skeptical, fearing that its downsides, particularly for workers, far outweigh any plausible benefits AI might bring about. Plausible arguments can be made for each of these positions, but whichever position ultimately prevails, it behooves us to adjust our time frames before declaring victory for one or the other. If history offers any useful lessons, we should be thinking in terms of decades rather than a year or two, as is generally the case with both “sides” today.

Why the dash of cold water? Ironically, it is because of the very real potential of AI to transform economic life and society more generally. As the great historian of technology Thomas P. Hughes observed long ago, “radical innovation inaugurates new systems.” And such systems take time to build. Employing AI effectively and efficiently at scale is hardly “plug and play,”

In order for radical innovations such as AI to bend the curve, changes are required in various realms– economic and business organization, labor relations, education, social structures, and cultural orientations. Such changes generally occur via trial-and-error, marked by two steps forward-one step back non-linear paths. The U.S. experience is illustrative in this regard.

Let’s start with the “computer revolution” of the 1970s and 1980s. Early on, many enthusiasts assumed that computerization–particularly the widespread dissemination of high-power desktop and personal computers—and the heavy investment in information technology more generally would quickly increase productivity and, in so doing, reshape the workplace and lead to sharp increases in GDP per capita.

Over time, however, economists found that this scenario did not play out. Rather, they discovered that despite the rise in computerization, there seemed to be little rise in measured productivity. This finding led the eminent economist, Robert Solow, to observe in 1987—the same year he won the Nobel Prize in Economics—that “[y]ou can see the computer age everywhere but in the productivity statistics.” This pithy construction helped to inspire interest in what later became known as the “productivity paradox,” a coinage generally associated with economist Erik Brynjolfsson, writing a few years later.

Eventually, of course, we did see gains in workplace productivity—in the 1990s and thereafter. But first jobs had to be reconceived, workers retrained, workplaces reconfigured, work flows recalibrated, and deliverables reimagined. Before society more generally would reap significant benefits from computerization, culture, institutions, and the political order would have to change too, leading to the relative shift in position and power of various institutions–most notably, those relating to education and training—as well as of various social groups and constellations.

Taking the longer view of things, the time lag between computerization–a disruptive “general-purpose” technology with broad applications–and significant productivity gains was relatively modest, amounting to a generation or so. By way of comparison, it took much longer for the steam engine to have a great impact on the U.S. manufacturing sector, although its impact in transportation came more quickly.

The first steam engines were employed for motive power in British North America in the mid-eighteenth century, well before the United States even existed. Steam engines were employed in a variety of production activities, milling especially, by the early nineteenth century, but did not become the dominant power source in American manufacturing until the 1870s and 1880s. It took the better part of a century, in other words, for manufacturers to reposition their operations (they were no longer dependent on locations near waterways), and to reorganize their workplaces, work processes, and workforces to accommodate steam power rather than water power.

The subsequent shift to another technological system, based on electrical power, also took decades to beat fruit in the form of measurable increases in productivity. Electricity and electric power had become viable possibilities in manufacturing by the mid-1880s, but as economic historian Paul A. David demonstrated in a famous paper in the American Economic Review, it was not until the 1920s that the necessary systemic changes were sufficiently disseminated to make a large difference on the factory floor.

General-purpose technologies based on steam, electricity, computers, and presumably AI are of monumental importance, and in the first three cases have proven so over time. But AI will require a good bit of runway before it transforms the workplace, the workforce, and society more broadly in similarly profound ways. Let’s give it some time before we assess the results of AI for better or worse. Check back in a decade or two.

The post AI: A Timely Reminder appeared first on CounterPunch.org.