The Singularity

October 22nd, 2016

The discussion on singularity this past meeting seemed to draw nice parallels with parts of Frankenstein by Mary Shelley. In the novel, although scientist Victor Frankenstein set out to push the boundaries of existing techniques to create something that could be viewed as beneficial to the human race (making revival possible), his creation strayed from this set purpose and instead took the form of something hideous and unexpected. Although a fictional work, the novel conveys a message and in some sense, a warning. Invention, by nature, constitute exploration into certain unknowns. This process of seeking to create or to discover thus includes elements of ambiguity and uncertainty. Specifically because of these elements, there is considerable risk involved in pursuing areas without a clear vision of where the “upper bound” or limitations lie. The concept of singularity is indeed valid, and upon some reflection, I think it is possible for the process to happen in a “sudden” moment.

One could easily find discoveries or products that came from accidental moments in the lab. For example, Wilson Greatbatch, inventor of the pacemaker, did not set out to create a relatively small device that could be used to regulate the pulse of a human heart. Instead, he was working on a heart-rhythm recorder. Due to a mistake, he assembled the recorder using a different resistor than what he intended. As a result, the device took on a much different nature, producing electric pulses rather than recording rhythms. Although this instance and many other similar stories (discovery of x-ray) present favorable representations of the results of accidents in labs, one could sense the magnitude of impacts of these inventions. By extension, if similar “mistakes” are overlooked in the process of building smarter and smarter computers, one could envision a point at which the process becomes sudden and no longer reversible.

One point that especially stuck out to me was the concept of “natural selection.” The concept of computers reproducing to produce a better final product seems nearly unfathomable. As inventors, humans are so used to be in the position of authority to decide the specific components and functions that computers would be capable of. To think of computers as self-directing entities able to select desirable traits based on their own discretions imply a certain autonomy that, for me, is nearly uncomfortable to imagine.

One Response to “The Singularity”

  1. Mike Smith said:

    While not complete reproduction, you might take a look at the idea of simulated annealing in computer chip design. The idea here is to take a very simple algorithm and be able to produce efficient, complex designs. The result was often better than a human could produce. And this isn’t recent work. For example, see

    https://app.cs.amherst.edu/~ccmcgeoch/cs34/papers/rutenbar.pdf

Leave a Reply