Presentations
A selection of sets of scripts and slides. Fortunately, I think we have plenty of time for questions.
Re-Reading the Limits of Linguistic Abstraction
First-year composition pedagogies attempting to respond to the crisis of textual proliferation prompted by Large Language Models (LLMs) fail to adequately address the circumstance “that words have a life of their own, radio-controlled by an automatic pilot called history,” in Charles Bernstein’s 1987 words: “To understand language as artificial intelligence is to conceptualize writing as a kind of psychic surgery.” Compositionists might productively imagine writing that, via Bernstein, “sounds as music the marks of its continual newness in being made; and the only mark of its past, of its having been made, is the blood on the cutting room floor.” Emerging technologies simultaneously obscure and reveal tensions between abstracted knowledge processes and material-historical human labor.
Questioning Generative AI: How to Not Write Like a Robot
Composition’s reception of generative AI has been shaped by a disciplinary wariness of numbers. That wariness costs the field an insight it already owned: Peter Elbow’s published characterization of the reading process describes the predictive machinery of a language model in composition’s vocabulary, with parallels easily traceable to the mathematical processes of stochastic gradient descent in Large Language Models (LLMs). I use Mariolina Salvatori’s diagnosis of the double erasure of the reading student and of that student’s reading processes to argue that LLMs are better understood as ‘two-player word processors’ than as producers of pre-completed and pre-plagiarized texts, and that their failures are precisely the sites of the appropriation of the value of students’ writing labor.
Owning Your Digital Shadow: Students’ Right to Their Own Data
In 2017, CCCC Chair Linda Adler-Kassner gestured toward the problem of educational paradata, the recorded “data exhaust” of student work that under US law belongs to the collector rather than the producer. Framing educational paradata as a technological by-product rather than an authored text makes it appropriable at the moment of creation. Adler-Kassner’s framing presumes the appropriation of value by parties other than the creator and blames the appropriation of value on the quantitative measurement of value, manifesting what Victor Villanueva has referred to as a humanist fear of numbers, and ignoring the productive uses of measurement as indices of the appropriation of value by the producer or by other parties. I offer a case for data sovereignty: the right to download, own, analyze, and argue from one’s own record of composing.