The necessity of a liberal arts background for STEM is made apparent nowhere moreso than AI research.
Because clearly they have not understood Mary Shelley’s Frankenstein.
You know, the one subtitled “The Modern Prometheus”?
It’s not “horror” in the modern sense. It’s horror in the sense of psychological dread, the unexamined consequences of one’s actions, and a creator’s responsibility toward the created.
Being conscious and aware is a mixed blessing at best. The joke in Rick and Morty’s “butter passing robot” is not that the self-aware robot is only there to pass butter.
It’s the last, less-quoted line from Rick: “Yeah, welcome to the club.”
That line matters. It shows Rick’s default way of interacting with the world is sociopathic (something too many fans forget), but that Rick also realizes the joke is on everyone, including him.
None of us know what the purpose of life is. It’s existential dread, and it can be devastating.
The horror of Dr. Frankenstein’s hubris is the same horror we laugh at in Rick and Morty. We ignore it in order to function; that’s why we laugh at the robot’s “Oh my god” instead of Rick’s “Welcome to the club.”
We talk about the utility, environmental costs, and copyright implications of LLMs. We absolutely should.
We do not talk about making sure that we are not cruel. (1)
Artificial consciousness is considered possible as an emergent property. There are researchers trying to create exactly that kind of strong AGI — artificial general intelligence — using current models of personality theory and psychology.
This isn’t to argue that LLMs are conscious and self-aware.
But why would you strive to create an entity that is capable of having an existential crisis? That is sociopathic cruelty.
The reasons folks engage in this kind of development and research come down to “understanding our own consciousness” or “gaining knowledge”.
We have ethics rules about human experimentation, both medical and psychological, developed in the 20th after abuses: Mengele, Tuskeegee, Zimbardo, Milgram, more. We still use data and information gathered in those experiments.
Those experiments gave us knowledge about the human condition (or gave the impression of doing so).
But they were also cruel.
So we stopped doing those experiments.
So when we find ourselves acting out a reverse Milgram experiment with LLMs, when we attempt to create something that could experience an existential crisis, the issue is not whether or not we succeed in doing so. The horror is not that it is created.
The horror is finding out that we are cruel enough to know exactly what we are doing… and we keep doing it anyway.
(1) We don’t guess if the anesthesia has taken effect before starting the surgery, why would we guess that we’re not being cruel? And before the vegans and vegetarians say it: Yes, I know.