Exploring the Potentials and Pitfalls of GPT Models in Scientific Research

Video recording:

Speaker: Raul Ian Sosa (Department of Physics; Faculty of Science Technology and Medicine, University of Luxembourg)
Title: Exploring the Potentials and Pitfalls of GPT Models in Scientific Research
Time: Wednesday, 2023.03.22, 10:00 a.m. (CET)
Place: fully virtual (contact Dr. Jakub Lengiewicz to register)
Format: 30 min. presentation + 30 min. discussion

Abstract: In recent months, GPT models have become increasingly popular as powerful tools for natural language processing tasks, such as summarization, language translation and text generation. In this presentation, we will provide an overview of GPT models, their architecture, and how they were trained. We will also discuss the potential applications of GPT models in scientific research, including their ability to generate hypotheses and assist with scientific writing. However, while GPT models have many promising uses, they also come with significant pitfalls, such as the potential to produce and propagate false statements. We will discuss these challenges, understand where they come from and explore ways to mitigate them in order to ensure that large natural language processing models of this kind can be used responsibly in scientific research.
* In compliance with current copyright laws we confirm that this abstract was generated by a human.

Raul Ian Sosa is a doctoral researcher in the Department of Physics and Materials Science, in the Faculty of Science Technology and Medicine of the University of Luxembourg.