The Northeast Association for Language Learning Technology (NEALLT) will hold its annual conference online on Saturday, April 1. The NEALLT ’23 conference will be hosted by the Swarthmore College Language Center.
This year’s theme is Growing Together in the Digital Age: Unlocking the Potential of Technology in Language Learning.
KEYNOTE
Timothy Laquintano
Lafayette College
Large Language Models:
What We Need to Know Before Our Students Adopt Them
Artificial intelligence engineers still have a number of significant issues to address before large language models like ChatGPT experience mass adoption. However, some early evidence suggests that multilingual learners will adopt LLMs in enormous numbers to write more like native speakers as they learn languages, seek higher education, and participate in the world of international business. What will it mean, then, to be assisted in writing, learning, and thinking by a machine that has been trained on heavily biased data—to the extent that the AI has adopted a “hegemonic worldview”? Big Tech has already begun human-based reinforcement learning to prevent LLMs from spewing racist and misogynistic language, but they have been reluctant to share the intricacies of these methods with researchers. This talk will argue that if we look at patterns of content moderation on social media sites over the last fifteen years, we will have a tentative understanding of the approach Big Tech might take to training the politics of AI. This will provide us with a more critical understanding of the tools that many of our students are poised to adopt to assist their learning experiences in higher education.
Timothy Laquintano holds a PhD in English with a specialization in writing studies from the University of Wisconsin-Madison. He is associate professor of English and director of the College Writing Program at Lafayette College. Laquintano is a qualitative internet researcher who studies how people adopt digital communication technologies. He’s completed studies of self-publishing authors who achieved bestseller status, elite poker players who wrote and monetized instructional ebooks, Twitter bots that circulated propaganda, and professional writers who worked in toxic social media environments. His book “Mass Authorship and the Rise of Self-Publishing” (U of Iowa Press) won the 2016 Computers and Writing Distinguished Book Award. Laquintano is currently at work on two projects: a qualitative study of writers who use large language models at work; and a co-edited collection of pedagogical resources that will help faculty teach with large language models.
For additional conference information, please contact Michael Jones (mjones1@swarthmore.edu), or Angelina Craig-Flórez (Angelina.CraigFlorez@columbia.edu)
We look forward to receiving your proposals!
Our Hosts at the Language Center at Swarthmore College include:
- Michael Jones, Director
- John Word, Technologist
