Moore's Law, Broadly
In this excerpt from "SIR: A Computer Program for Semantic Information Retrieval" by Bertram Raphael and edited by Marvin Minsky in the book Semantic Information Processing, we are offered several suggestions of areas where computers being able to really "understand" would be helpful. A modern listener can listen to these few paragraphs, nodding along. Information retrieval, more specifically document and fact retrieval, and mechanical translation, are all problems that modern computer scientists still work on. Rigid databases are giving way to NoSQL databases and other more flexible data sinks. Google continually updates its algorithms, as do other search sites; Watson is out there beating us at Jeopardy; retrieval from the multitude of nodes in the Internet is a tall order. Translation is still not terribly fluid, though a number of apps are getting closer. These two examples, certainly, are areas that we have made tremendous progress in in recent years -- but areas that we still definitely have a long ways to go.
It can be somewhat disorienting, then, to realize that this excerpt is from a book published in 1968. These same problems, identified half a century ago, are still among the main problems worked on with machine learning and other AI systems that we're seeking to build. For sure we've made so much progress, but half a century ago would anyone have expected these "problems" to take so long to solve? It reminds me of an xkcd comic; we keep coming up with new problems to solve -- many get solved, and many that seem insolvable end up getting connected back to problems we already were struggling with... like search and translation. Could those end up being the points that need to be overcome for a machine intelligence to really pass the Turing Test?
We've become inured to the pace of progress. Many of the new developments taking place practically daily just blow right by us; we're swamped, there are only so many things we can really take in at a time. Take someone from just fifteen years ago, and place a smartphone in their hands. It's really hard to believe smartphones have really only been around for about a decade. They just seem like a perfectly normal part of life to those of us lucky enough to live in developed parts of the world.
Reading computer science texts from the 60s and 70s are frequently eye-opening. I was lucky enough to inherit a number of such books, and it's always interesting to hear what a previous generation had to say about the new digital world they were only then beginning to build. But much like our predecessors, we might be fascinated by things that will seem utterly pointless to some in the future. We'll convince ourselves that we can plot out where we want our technology to go, and somehow things end up heading in a different direction.