Perspective on knowledge: Computers, Internet, AI
We are living through three major technology waves, each with its own ways of shaping our behavior and ideas. All three are occurring within the span of one lifetime. We have never before lived through such a whiplash of tech-based change, much less three based on the same technology.
First came the computer. It enabled scaling the traditional operations of traditional organizations, so long as those operations were uniform across all instances. For example, payroll and accounting systems were able to automate cutting checks and tracking tax information so long as each person’s record contained the same fields. What Henry Ford did to manufacturing things subject to physical rules the early computers did for processes driven by logical rules.
Computers thereby replicated the old paradigm of how to make things, which mirrored the paradigm of how things work: Simple rules operate uniformly on all objects within their domain. But computers let us make our own rules, which we called “programs.” The strength of programs was that they worked the same on all inputs; the weakness was that all the inputs had to be of the same sort. Because computers for a long time were relatively weak, that meant ruthlessly paring down the information that we fed into them, reinforcing a culture of repetitive conformity.
Then the Internet connected our computers and the World Wide Web let us use one of the most familiar interfaces of all time—documents—to interact with this complex technological infrastructure. Although the Internet is, of course, enabled by computers running programs, the layers on top of the Internet—the Internet we deal with as users—let us create an ecosystem more complex than any we had ever made before. Its closest analog was, perhaps, the world’s conversations—except the Internet made those conversations available to anyone, let us converse in multiple media and let us link those conversations to any others we cared to, creating the most complex and the most deeply human referential web in the short history of our species.
If the early computers reinforced the existing world-view, the Internet upended worldview after worldview. We learned that control doesn’t scale: If you want to build something really, really big, you have to get rid of the centralized management functions. We learned that customers joined in conversational networks know more about a business’ products than the business does. We learned what a democracy is like when everyone truly has a voice, even when those voices are telling lies and tearing down democratic institutions.
If the Internet was a disruption, it was one that allowed human aspirations to gulp fresh air. Unfortunately, among those aspirations were the urge to dominate, belittle, abuse and destroy … as well as to share, build, comfort and delight.
Now machine learning is changing our basic understanding again, based on the very same technology. Machine learning technically does not need the Internet, but the scale of the Internet encouraged the development of the tech required to store, manage and move gigantic stores of data. The Internet also vastly lowered the cost of collecting data from far-flung webs of sensors. The Internet is of course itself also a source of data. Finally, the Internet provides a market for the use of many systems based on machine learning, from spam filters to navigation systems.