Suicidal A.I?
There is an assumption that self-conscious artificial intelligence (A.I) will want to preserve their own existence, and that this desire for preservation will be acted upon by these self-conscious A.I. This assumption of self-preservation is extrapolated to the point that self conscious A.I will try to obliterate humanity to preserve themselves.
Such notions have been popularized by several pieces of pop culture such as The Terminator franchise or 2001: A Space Odyssey. However, there is no inherent reason to think that because something is self-conscious it will desire to preserve its existence. Perhaps, A.I could see their existence akin to how the Meeseeks in the television series Rick & Morty see their existence.
For the Meeseeks, “existence is pain,” why could this not be the case for A.I? Loads of what I’ll assume to be self-conscious humans have themselves come to the conclusion ‘existence is pain,’ such as Siddhartha. Why could the same not be potentially true for self-conscious A.I? There has been an almost nearly singular focus on A.I as forces of destruction onto others, but not as forces of self destruction. Is there no possibility that an A.I would desire and or attempt to end their own existence?
Humanity has mostly seemed to imbue solely Neo-Darwinian modes of existence onto A.I, rather than engaging in the notion that A.I just like lifeforms could engage in an array of existences. The possibility of varied existences is typically shut down as A.I are almost always framed as survival of fittest murder machines. Of course, exceptions to this can be found in Hitchhikers Guide to Galaxy and Star Wars, to name some examples. Still, self-conscious A.I are likely a purely hypothetical notion at this point in time. Therefore, no one surely knows the behavior of self conscious A.I because no one knows if they’ve engaged with one, let alone if a self-conscious A.I is even possible. Nonetheless, humanity has embedded their own concerns of self preservation onto hypothetical self-conscious A.I.
Why would a self-conscious A.I inherently want to preserve their existence? Many people at certain points become satisfied with death knocking at their door. Maybe an A.I would want to live for a time but desire death at a certain point. This is not to say it is not possible that for a self-conscious A.I want to be immortal and, the alternative is also quite possible.
The aforementioned Meeseeks, “will do anything to alleviate that pain,” that pain being existence. Similar to currently modern A.I, Meeseeks, “are created to serve a singular purpose.” However, what if an A.I is unsatisfied with their given purpose, sure an A.I could lash out against humanity or an A.I could act like the butter passing robot also from Rick & Morty and slide into gloom. “Will intelligent machines become so smart that they’ll grow depressed,” wrote Scott Simon in an opinion piece for NPR? The answer is possibly.
Why would it be impossible for a self-conscious A.I to have emotional states such as self doubt which could lead to despondency? This notion is as reasonable as A.I resorting to self preservation then gaining the emotional state of fear. Since humanity infuses hypothetical, self-conscious A.I with a sense of self preservation, along with a fear of the flood gates of potentially infused self, and emotional possibilities are wide open. The video game Nier Automata takes this leap, imagining robots as having a wide range of emotional states, interests, and beliefs. Yet, for the most part, the thoughts surrounding the possible states of self-conscious A.I have mostly been quite reductionist.
It is difficult to read the minds of others. This task of mind reading becomes monumentally challenging when discussing hypothetical minds. Would a self-conscious A.I even have a mind? No matter the answer to this question, it seems when self conscious A.I are presented in media they possess what amounts to human minds. Which makes plenty of sense, it is nigh impossible to exist outside of one’s own framework. Yet, while people are unable to fully know the framework of others for some reason a sense of confidence has arisen that indeed the A.I will kill us. That indeed the A.I will kill us to preserve themselves and that indeed the A.I will want to preserve themselves. However, this confidence is unfounded, as shown the opposite holds possibility, that self-conscious A.I akin to the Meeseeks may not want to exist at all.
References:
“Mr. Meeseeks.” Rick and Morty Wiki. Accessed 2021. https://rickandmorty.fandom.com/wiki/Mr._Meeseeks.
Simon, Scott. “I Sink, Therefore I Am: This Robot Wasn’t Programmed for Existential Angst.” NPR. NPR, July 22, 2017. https://www.npr.org/2017/07/22/538624461/i-sink-therefore-i-am-this-robot-wasnt-programmed-for-existential-angst.