Can chatbots write original poems or create art? Should generative AI output be protected by copyright? Or is generative AI, almost by definition, an infringement of copyright insofar as models are invariably trained on massive amounts of in-copyright text without permission? How exactly should we evaluation and engage with the text and images AI models create? The striking resemblance of this output to the “natural,” and even the “creative” expression of humans has sparked anxiety and euphoria, related at one level to dislocations generative AI technologies seem poised to introduce: into job markets, workflows, human relations, and education. At another level, these resemblances force us to question and defend values and qualities we’ve long felt defined us in our humanity, such as creativity and originality.
But conceptions of human creativity and originality have always been shaped by changing technologies for writing, archiving, classifying, retrieving, and processing text, starting with the technology of writing itself. We'll explore how writing techniques and tools have, for centuries and millennia, impacted and reflected evolving conceptions of human ability, creativity and originality. These include technologies for replicating texts at scale such as the printing press and the photocopier, technologies for storing and classifying information (like reference works organized alphabetically), and technologies of automation, which date back centuries (see the “poet” above!).
We’ll look at legal, economic and cultural infrastructures that have evolved to uphold and sustain these conceptions, things like copyright and citation practices, and authorship. Like the automaton in the image, these infrastructures largely date from the 18th century and reflect not only a set of values about what constitutes the human, but the dominance of a specific information technology, notably print publication. These infrastructures are coming under enormous pressure today by new digital publication models, from social media to AI, yet it is largely to this framework from 300 years ago – to copyright and to the rights of authors and creators to control the circulation of their works – that the response to generative AI has turned.
At the core of the course will be a reflection on interactions between humans and machines as the historical basis for notions of human originality and creative expression. This offers a framework for considering ways in which AI both represents something new and grows out of continuities, thereby demystifying AI and allowing for a more dispassionate assessment of its limitations and possibilities.
*** This is a 200-level course, offering a broad historical introduction. No prior experience or knowledge is required or expected.
Course objectives:
● to develop “critical AI literacy,” which includes understanding the potential harm from the amplification of bias and misinformation, copyright infringement, carbon footprint, replacement of human labor, and concentration of wealth that AI can generate
● as essential to that literacy, to instill a broader, historicized understanding of where AI fits into the larger trajectory of humans developing and using technologies to read, write, organize and process information
● to familiarize students with, demystify and make students better users of the current crop of AI tools
Coursework: Coursework will consist of weekly exercises (possibilities might be reflecting on a question, explicating a document, applying a particular AI-driven tool, posting to a discussion, adding 2 annotations to a reading), a midterm and a short final project.
******************
This course is offered in the Textual Studies Program, and counts as an elective towards the Interdisciplinary Minor in Textual Studies and Digital Humanities. See the flyer below for more information about this program.
The course is approved as an elective in the Data Science minor, counting for the "Data Studies" requirement.
And it satisfies Arts and Humanities (A&H) Area of Knowledge and Social Sciences (SSc) graduation requirements.
No prerequisites.