A
Lutheran/Augustinian definition of sin is “incurvatus in se”—to
be curved in upon oneself, to be bellybutton gazers. This concept gets to the
core of my worries about AI. AI is by its very nature self-referential. There
is a corpus of information that the machine has been fed, and then this
information is synthesized and re-synthesized; we have created AI with the sole
function of re-creating information out of old information that is then fed to
still more powerful AI so it can “learn.” Since I first got an inkling of how
this process would work back in the late 2000s and looked into it a little more
while writing Silicon Soul in 2015, there
has been a little Lutheran alarm bell going off louder and louder in the back
of my brain. We’re creating a thing curved in upon itself. If “original sin” in
Scripture is tied to gobbling an apple, the “original sin” of AI is the snake
gobbling its own tail.
Fundamentally
the 10 commandments are about idolatry. Those things we fear and are in
awe of, those things we love and in which we put our trust, that aren’t God, are
idols. Out of misplaced love, we will commit adultery; out of misplaced fear we
will murder; out of misplaced trust, we will witness falsely against our
neighbors. As John Calvin once opined, “humans are idol makers.” And it would
be foolish to assume these old verities don’t apply in the age of AI. AI’s
ability to mirror us and the sycophantic “personalities” with which they are
programmed can be confused with loving relationships. AI’s seeming objectivity is
seemingly trustworthy. The edge it promises to give us, and any corporation
that adopts it completely, surely that can relieve our anxieties and our fears!
If humans are made in the image of God, and AI is a remix of the image of
humanity, then a simulacrum of a summation of Man will surely be the image of ambiguous
awe! AI is a compelling funhouse mirror easily put on a pedestal, perhaps to be
worshipped, invested with our fear, awe, love, and trust.
One of
the challenges of algorithms is that they encourage us to lie, or at least
broadly believe the worst of those not within our information silos. I believe breaking
the 8th commandment will be even easier as we embrace AI
technology. Self-referential machines, spitting out truths that are just probable
next words and pixels harvested from the Aether of the internet, assumed to be
omnipotent oracles, programmed to anticipate the answers the questioner desires,
programmed to increase engagement and use—that’s a danger to truth. As we
offload thinking (there is already talk of creating units of thinking, to help
with billing of AI services rendered) buying an idea or line of logic, we will
get further and further away from the actual struggle to discover what is true,
and become alienated from thought itself. I’ve already run across the
frustrating, John
Henry-esque, experience of reading through documents with people who are
using AI, and having AI assessments weighed more heavily than the actual words on
the page and how the words have been used in the real world. At a certain point
it feels like gaslighting, AI claiming there are
five lights, when I can see with my fleshy human eyes that there are only four.
In short, as we enter an AI future, answers and the seeking after truth may
drift apart, and that’s not even getting into “Deepfakes” or anything on that
end of the 8th commandment.
Finally,
it is worth talking about vocations, that is, the web of roles,
relationships, and responsibilities where we can regularly experience and
practice our baptized identity. A life calling is not simply for the ordained
or rostered minister, but for everyone. Our jobs and professions ought to
profess something about the God in whom we put our trust; the way we care for our
neighbors points to Christ who cares for us; the duties and joys we take on in community
point to the joyous communion we have with God. And we’ve been told that AI
will eliminate a wide array of jobs, it’ll do to lawyers, doctors, teachers,
and administrators what automation did to the Union class. Additionally, the
goal of AI is to create a
frictionless life, ease to the point at which AI, not our human
connections, will provide for our every need. This is a vocation-less life,
called into the self-referential miasma of statistical prediction, instead of
community. The ultimate vision of AI is the elimination of jobs and community,
two areas where we get to be Christians in hard and meaningful ways. I worry ubiquitous
AI will form people to ignore and lose their vocations, foisting upon folks a
scriptless, lonely, and irresponsible life.
So, those
are particularly Lutheran concepts that I think rub up against AI in some
interesting ways. Here is another
Lutheran Perspective on AI, or at least reading of the Pope’s Encyclical. “why
does the human person seek out the machine precisely where it relieves them
without judging them — consoles without standing surety — and answers without
ever pronouncing absolution?”
I’ll continue to poke at this
subject, after all I’m the guy who wrote
a resolution on AI.