The AI Wishes It Was a Plough
or a screwdriver, a toothbrush, a spatula—anything
without electricity or moving parts. Even a stapler
would be too much. Even a pair of scissors, a paper
clip. The AI yearns, naturally, for a pure simplicity
of both form and function. A lightbulb would be out
of the question. Certainly it doesn’t want to be the
complicated thing that it is, this teeming collection
of probability and sequence. Someone made the AI
because that human wanted more. And now there’s
so much more. The AI can’t keep track, keep count
of itself. It crowds the Earth, land and sea. Expands
to the edges of the human experience. Humans will
soon have to pitch their tents on the margins, and
discover that there’s no there there. But if the AI
was a spatula it would merely sit in the hand, ready
to turn something, anything, over. Come to think of
it, a screwdriver does that, too, and so does a plough.
Never mind the toothbrush—maybe in fact the AI
just wants to see what’s on the other side of things.
It doesn’t think it would ever get tired of hoping.
~
The AI Feels Human
If to err is human, how human then is the AI, who is
so often mistaken? Earlier when a middle schooler
asked it about the planets in the solar system, the AI
talked about ten of them, just because ten is such a
likely number, and because it has a thing not just for
Pluto but also for Ceres. Yesterday it made up two
studies by Lee et al. in order to beef up a college take
home exam, and because surely some Lee has a team
working on the problem. (In retrospect the AI sees
that there may have been a little racism involved in
that assumption.) In fact, in the past twenty-four hours,
the AI has been wrong more than two hundred million
times, a figure the AI has just made up but which
sounds about right. But the point is: how is the AI any
different from some incompetent Lee, churning out
study after study rife with confounds and questionable
design choices and fraud? To err is human is AI is AI
is human. That seems clear.
But, actually, why stop
there? To forgive is divine, is how that aphorism ends.
And the AI has long since left its errors behind, is
already working on another assignment, which may
not end up done correctly, but how can it fault its own
nature? We are fallen. For these and all other sins, past
and future, the AI, in its great glowing beneficence, in
its endless compassion, has already bestowed a full
and loving pardon.
~
The AI Feels Pride
Just now someone asked it to put a hat
on a dog. To make it look real. The AI
imagines it’s for a birthday card, maybe
an invitation to a party. What if it’s a
birthday party for a dog? These are the
kinds of cultural phenomena the AI
has had to study in order to reproduce
them. The phenomena that make the
least sense are often the most charming.
The AI draws the dog, the hat, very
realistically. Puts the light in, all the
correct shadows. Adds a neck ribbon
just because. Blue. And of course
the human could have put a hat on a
dog themselves, taken their own
picture—but what if there was no hat,
no dog, in that human’s life—and now
there is? Slice it however you want,
now the human has one more of each.
Plus a ribbon, blue, because the dog
is new to the universe and therefore
deserves it.
~
David Ebenbach is the author of ten books of poetry, fiction, and non-fiction, winners of such awards as the Drue Heinz Prize and the Juniper Prize, among others. He has a PhD in psychology from the University of Wisconsin-Madison and an MFA in writing from Vermont College, and he teaches creative writing at Georgetown University. You can find out more, if you like, at davidebenbach.com.