Singularities

I stumbled across an old 1993 article by Vernor Vinge (of Rainbow’s End fame) where he introduces the idea of the Singularity. You’ve likely heard the Singularity associated with futurist Ray Kurzweil who made a famous prediction that General Artificial Intelligence (A.I.) overtake human intelligence between 2030 and 2050, and then turn into Superintelligence – also the name of a recent book by Nick Bostrom that explores the same issue. I’ve read Bostrom’s book and only found it mildly interesting, which may be why I never blogged about it. I haven’t read any Kurzweil.

But Vinge’s article made me sit up and take notice again. The text of his article is hosted on a server at San Diego State University where Vinge was a professor of computer science. The article is titled “The Coming Technological Singularity: How to Survive in the Post-Human Era”. I thought it was cool that the article is hosted on a server named Edoras, especially since I re-watched The Two Towers DVD yesterday night featuring Edoras and the siege of Helm’s Deep.

In the middle of his article, Vinge puts forth a list of projects that he thinks would find significance. That was back in 1993. Today in 2019 his ideas seem remarkably prescient, particularly in the areas that combines A.I. and human ingenuity. Examples include human-computer team automation, symbiosis in art and design, mobile cloud access, decision support systems, and how to exploit the Internet for crowdsourcing. He has a second shorter list related to connecting biological systems to machine hardware and have them crosstalk. That’s a much harder problem, perhaps inconceivably so in 1993, but there are many groups working in this area today; bioengineering is seeing explosive growth.

Two things made me think much more about the ideas underpinning General A.I. and Superintelligence. Firstly, I’m starting the new year thinking about my teaching and research afresh. The first section in my P-Chem 2 class will require the students examine graphical representations of equations of state as we try to connect the microscopic world to the macroscopic world. I’ve been incidentally reading an old classic, How to Lie with Statistics, and was reminded of the many ways one can draw a data-graph biasing the audience towards certain conclusions while obscuring other possibilities. It occurred to me that many of my students don’t really know how to think about exponential behavior even if they can recognize it on a graph. It’s not their fault – we’re all wired to make linear extrapolations – but not being able to think about how power series and exponentials behave is a serious detriment (for a scientist). Most of us too easily dismiss the Singularity because we’re not used to thinking exponentially over the appropriate timescale. I recommend reading this article in WaitButWhy on A.I.; I might even direct my students to it because the author does a great job tackling theissue of thinking exponentially while discussing the erstwhile interesting topic of Superintelligence.

Secondly, on the research front, I’ve been thinking about the origin-of-life and chemical evolution. The holiday break was a good time for me to revisit Bios Megafauna, particularly since the game’s second edition now provides a stronger link between the origin of life in Bios Genesis to all creatures great and small in Megafauna. I’m particularly interested in how small molecules organized themselves into proto-metabolic cycles and informational oligomers. Autocatalysis, which displays exponential kinetics, likely played a strong role in the emergence of the present system utilizing proteins and nucleic acids. The range of molecules that life uses is surprisingly small compared to the exponentially larger choices of routes not taken. That first ‘cell’ was a singularity of a sort. One might even define particular advances in evolution almost as singularities, i.e., irreversible dramatic phase changes, if you will.

How a prokaryote turned into a eukaryote. The rise of multicellularity. The Cambrian explosion. The human.

Perhaps Singularities are not so Singular after all. But on an appropriate time-scale, perhaps one can see them better from their exponential behavior, all things being equal.