Originality has long been the last bastion of human creativity. Sure, John Henry may have died in the end and lost to the locomotive, but human beings have generally taken pride in the steadfast certainty that even when Skynet does eventually take over, at least the damn machines won’t be able to paint great art, compose orchestral masterpieces, or write works of literary genius that touch the deepest depths of the human heart and soul.
Well, no. At least, I don’t think so.
Bagel asked me yesterday why exactly I was embarking upon this crusade to input an entry a day, cycling through the alphabet as many times as necessary, in order to put everything I know, each entry 300-700 words, into WordPress. Aside from being a fun exercise that helps me practice writing daily, the other real reason is I’m trying to generate a corpus of material for Wobble2– a facsimile of all of my thoughts, positions, opinions, and beliefs.
I have a theory, entirely unproven, that given the right corpus, I could code a reasonable replica of myself, at least for a limited universe of Q&A. The dream here is to write software that could eventually synthesize answers in a way that I, Wobble Prime, would answer them. This premise is directly inspired by The Turing Test and TV shows like Westworld and Devs. If the simulation generates identical answers as the original, why is the simulation any different or “less than” the original? (Gotta love humans and our fascination with making ourselves obsolete!) So, here’s an example desired outcome: Eventually, I want to build out Wobble2 to a state where it could answer a question like, “Would Tom Hanks make a good president?”
To answer this question, Wobble2 would need to look up its entry for “Tom Hanks.” And then it would need to look up its entry for “president” (and infer that the question meant, “President of the United States”). From here, Wobble2 would need to compare all of the qualities I associate with good/bad presidents and cross-reference those traits with Tom Hanks’s characteristics. Finally, depending on the results of that comparison, Wobble2 would then render a “Yes/No” response, an emotion associated with that response (“hell yeah!” or “tepidly optimistic?” or “that’s a very bad idea.”), and some supporting body of evidence for how it got to its conclusion. I still have kinks to work out, but in my head I think it could work! This is, in a nutshell, how I’ve been spending my time these days.