Kevin Stowe

News

Work

  • I'm currently a Research Scientist, NLP at ETS, working with the Automatic Content Generation team to build better education and testing tools!
  • Previously, I worked as a postdoctoral researcher with Prof. Dr. Iryna Gurevych at the UKP Lab in Darmstadt, Germany.
  • Before that I was a research assistant at the CLEAR Lab at the University of Colorado under Prof. Martha Palmer while pursuing my PhD. I also worked for Project EPIC under Prof. Leysia Palen.

Education

  • I completed my PhD. from the University of Colorado in August 2019.
  • Masters in Linguistics from Indiana University, Dec 2010.
  • BA in Linguistics from Michigan State University, May 2009

Interests

  • Responsible AI: How can we ensure LLMs and other language technology is safe, fair, equitable, and sustainable? I'm particularly interested in issues surrounding safety, accessibility, and societal/environmental impacts.
  • Language Generation: Using language models (including but not limited to GPT) to generate text is fun, and getting better: how can we control the outputs to ensure they match our requirements while mitigating bias, fairness, and hallucination issues?
  • Computational creativity: How do machines process creative language? Can we detect it? Can we generate it? Metaphors, metonymy, sarcasm, humor, and more all pose significant challenges even to state-of-the-art systems.
  • Crisis Informatics: How do we best make of use of the massive amount of data created during crisis events (including natural disasters like hurricanes, terrorist events/wars, and pandemics like COVID-19)? Social media can help, but applying NLP can be challenging due to privacy, safety, and quality concerns.
  • NLP for Social Good: Relatedly, how can we leverage the vast amount of power AI has to improve people's lives in concrete ways? How do we deal with ethical issues, problems of inclusivity, and systematic bias in data, systems, and researchers?
  • Language: My background is originally in linguistics - I'm always interested in studying linguistic problems, particularly in semantics and pragmatics. Other areas of interest include construction grammar, frame semantics, and creative language.

Repositories

I've contributed code for a number of research-based repositories around natural language processing, machine learning, data collection/annotation, and LLM-based methods.
  • (Coming Soon) Fairness Detection for Automatically Generated Educational Content
  • Controlled Language Generation for Educational Assessments
  • Generating Conceptual Metaphors with BART
  • IMPLI: Investigating Natural Language Inference Relations for Idioms and Metaphors
  • Exploring Metaphoric Paraphrase Generation