This panel will showcase the interdisciplinary projects of the three first-year postdocs in the Center for Science and Society:
Modern psychoanalysis and the creative self
In the earliest formulations of psychoanalysis, the analyst would “stand back” and “dig down” into the patient’s past and present, using interpretation as the primary tool to promote therapeutic change. This apparently detached stance gave rise to the image of the psychoanalyst as an observing blank slate. Successive generations of psychoanalytic thinking have challenged this paradigm, and today, some contemporary psychoanalysts see the analyst and patient as co-participants in a mutually affecting process of transformation, and as co-authors of a new narrative of the patient’s self. This new school of psychoanalysis leverages the creativity inherent in our capacity to develop a sense of self out of the complexity of experience, bringing psychoanalytic practice much closer to the creative arts. In this talk, I consider this recent revolution in psychoanalytic thinking alongside the rise of the literary genre of fictional autobiography, and suggest that evidence from research psychology and brain science supports the effectiveness of this paradigm shift.
Vox Ex Machina: Language Processing in Humans and Machines
Natural Language Processing, the domain of computer science dedicated to the creation of algorithms that can parse and generate text, has made tremendous progress in recent years. Last May, OpenAI unveiled GPT-3, the latest development in this line of research. GPT-3 is capable of producing whole paragraphs that are not only grammatically coherent, but well-written and topically relevant - so much so that they can fool people into thinking a human wrote them. However, algorithms like GPT-3 lack perceptual knowledge about the world, as well as beliefs, goals and values. In what sense can we say that their outputs are meaningful at all? And what are the ethical implications of our answer to this question for the practical uses of these algorithms? In this presentation, I will present some reflections on these questions.
“As close to robot as possible”: Modeling students through predictive analytics in higher education
Predictive modeling projects in higher education in the U.S. have contributed to a proliferation of student data, which universities increasingly use to predict and intervene on graduation and retention outcomes. Some of these interventions are student-facing. Data visualizations and nudging, for example, prompt students to be more productive and better manage their time. Informed by ethnographic research on the development and deployment of predictive modeling at a large public university in the U.S., in this talk I ask what kinds of data subjects are produced through modeling efforts. In particular, I focus on the rendering of students as poised to become, as one student put it, “as close to robot as possible” through the modeling of what students could be if they acted on their data.
Talks in this series will be followed by discussion, including a Q&A session with the audience.
Please email [email protected] to request disability accommodations. Advance notice is necessary to arrange for some accessibility needs.