Community Corner

Alicia Hofelich Mohr, Ph.D.

This month, the DSAI Community Corner spotlights Alicia Hofelich Mohr, Ph.D., Research Data and Support Services Lead in the College of Liberal Arts LATIS.

Please join us in celebrating Dr. Alicia Hofelich Mohr's innovative contributions to AI and data science, read her answers below!

What are your current research interests, and how does AI intersect with your work?

My current research and professional interests are around open science and data sharing - how do we make sure our research workflows are reproducible and how do we make our data available so that others can verify our work and reuse the information we collect to contribute to new scientific discoveries? I'm interested in how AI can streamline this work, but I equally share concerns about unexamined biases that might creep in as we rely more and more on AI in our workflows. AI models and users also benefit from data sharing, especially from high quality datasets being available, well curated, and machine actionable. As data creators and curators, we can make more reproducible and scientifically sound data sets available to AI models to hopefully help AI outputs be in turn more accurate and scientifically sound.

How do you define Data Science, especially in the context of AI and machine learning?

I see data science as a very broad topic that seeks to expand knowledge by centering the examination, combination, and transformation of digital primary source material. Often this is quantitative data, but could also be text corpora, images, audio, or other media. While technical and statistical skills are central to the innovation of how these sources get combined or analyzed, disciplinary experts in the primary materials are critical for understanding and interpreting what comes out of it. Data science is meaningless if it happens in a vacuum, but can be a powerful tool in combination with deep expertise on the primary materials. This is why I think universities are critical voices in the AI and data science landscape. 

Can you share an interesting or surprising result you’ve found in your data?

One of the most exciting things found in my data was not something we were looking for, but rather the potential someone else found in our data. Several years ago, my colleagues and I did a study on how the visual design of survey response boxes influences how people respond to questions. We shared all the data and code in DRUM when we published the article, and didn't think much more about it. A few years later, I was contacted by a research group who were interested in using our data to help develop machine learning algorithms to score one of the tests used in our study. Scoring of this test was traditionally done manually by several raters, which takes a lot of time and person hours. It turned out the data we shared as part of our study was one of the largest available datasets of how people score this data. They were able to use it for their model and they cited our dataset when they published their paper, which was completely different from our original use of it.

Are there any interesting new Data Science or AI tools, models, or libraries you or your students have been using in your work?

I'm still exploring all the different tools out there for AI, and I feel like I still have a lot to learn. I have enjoyed exploring notebook LM, and see a lot of potential in creating curated sets of information I can use AI to query. I also know many of my coworkers have been using AI tools to help refine code and automate different tasks, which have brought both great wins and head-scratching misses. 

What are you most excited about in your field with regards to Data Science and AI in the next 5 years?

Maybe it's my liberal arts background, but honestly I'm most excited about the opportunities for centering the human in all of this. With the accelerated development of sophisticated technical tools we need digital literacy and critical thinking skills more than ever. The way we've thought about scholarly work, intellectual contributions, data, and how we generate knowledge are all being changed by these tools. We need to think about our roles in this changing landscape. I'm also eager to see how AI can help us connect pieces of a workflow to make it more streamlined and our results reach more people. There's a lot of great work happening around machine-actionable data management plans, for example, that could connect large pieces of the research workflow, connect researchers to University infrastructure, support, and compliance help when they need it, and help make all the pieces of research more findable.