According to Elizabeth Callaway, assistant professor of English at the University of Utah and affiliated faculty
with the Environmental Humanities Graduate Program, the problem of artificial intelligence
is not just a scientific or engineering problem, it’s a cultural problem. And it’s
one that has escaped criticism for far too long. Her upcoming book is an exploration
of AI and the nonhuman environment, but it started out as something quite different.
The seedling for the project began, like so many projects, with a deep personal interest.
Callaway became fascinated – and horrified – by social media recommendation algorithms
that fulfill their mission of keeping users on platforms by sequentially recommending
slightly more extreme content. The result is polarization and radicalization of individuals
and society. “AI,” she says, “has been affecting us on social media without us realizing
it for a long time.” Initially, Callaway thought this was a personal interest, one
that emerged from thinking about screens and the developing brain in relation to her
own family, but she realized that AI actually touches every aspect of our lives and
that AI extends beyond the human to affect the nonhuman world. Running AI, for example,
is extremely energy intensive with a notoriously large carbon and water footprint.
This realization also came paired with another realization that there is less
attention being paid to what AI is doing to animals, plants, and ecosystems, a kind
of gap in the cultural criticism available – one that Callaway could at least begin
to fill with her research. In all aspects of her research and scholarship, Callaway
sees what she does as bringing science into conversation with the humanities. This
is certainly the case for this current book as her research includes real-world applications
of AI as well as literary portrayals of AI in novels, video games and children’s fiction.
Callaway draws from a wide variety of texts because not all impactful cultural portrayals
will come from traditionally “literary” sources. Science fiction novels and games,
for example, “excel at making our thinking more visible to ourselves, which can give
insight to why we’re doing the things we’re doing with AI in the real world. In addition,
these imaginative, fictional texts can function as thought experiments delimiting
the possibilities of how we think about AI.” Conversely, the engineers in the real
world working on AI often write about it in science fictional terms. Paying attention
to these science fiction narratives might help to situate a better, more thorough
understanding of AI.
But AI is not just the havoc-reeking technology she expected. In addition to the
many ways people are trying to use AI to protect the natural world, Callaway was pleasantly
surprised (but also somewhat unnerved) to find certain types of AI rather delightful
to use. There is Midjourney, for example, an AI that creates completely novel visual
art that is sometimes astonishingly beautiful and sometimes hilariously strange. There’s
GPT-3, an AI that will generate a surprisingly cogent if completely generic piece
of writing when given a prompt by a user. And there’s Quick, Draw! which uses a neural
network to guess what you’re drawing in 20 seconds or less.
Many of these instances of AI were shared with Callaway by her colleagues at the
National Humanities Center Responsible AI Curriculum Development Program – a symposium
she participated in during the summer of 2022. Part of the program includes an award
to develop a course in Responsible AI, which she is slated to teach during the school
year 2023-2024. The course, to be taught through the English Department as “Digital
Cultures: Responsible AI” is an example of how Callaway sees her research as inseparable
from her teaching. She views this course as an opportunity to share something she’s
really excited about with her students. With AI in particular, she also sees it as
an opportunity to reach students across campus in different disciplines and to teach
students outside the humanities to dwell in these complicated ethical questions.
Ultimately, Callaway sees this book project and course as using literature to
address a large problem in the real world. She says, “AI is often sold to us as the
answer, the more objective way to do things because a computer is used. People imagine
that AI eliminates human bias, but it doesn’t. It is created by human beings, and
it’s trained on the data we generate. Famous examples like Tay, Microsoft’s racist
twitterbot, show us the extent to which biases are baked into AI.” And because of
this, there desperately needs to be more criticism from every discipline to address
issues of AI.
Jana Cunningham, University of Utah College of Humanities
email@example.com | 801-213-0866