Welcome to "How I AI"—where innovation meets education.
In this inaugural article of "How I AI", Vice Provost and Graduate Dean, Prof. Hrant Hratchian, shares his insights into AI’s role in academia, bridging the theoretical and practical. From coding assistance to academic writing refinement, he discusses how AI can augment, but not replace, the creative and intellectual processes. His nuanced approach—acknowledging the challenges while championing thoughtful adoption—provides a blueprint for how institutions can navigate the evolving AI landscape.--- Interviewed by Sayantani Ghosh.Photo: Hrant at work, as visualized in ChatGPT 4.0
In his administrative capacity Hrant sees his primary responsibility as one to provide opportunity and oversight to the graduate education mission and postdoctoral training mission of the university, and to serve as an advocate and representative for the university in that capacity.
But he is also a faculty in the department of Chemistry and Biochemistry, with a very active research group comprising six PhD students. In that capacity he is well-placed to discuss not just the big picture, but the details of everyday use of AI tools in research and academic writing.
Coding with ChatGPT
When it comes to integrating AI, Hrant got in at the ground floor. He was introduced to ChatGPT at a meeting called by the Council of Graduate Schools in December 2022. He remains a big fan of ChatGPT, referring to it fondly as ‘chat’. Since 2022, Hrant has used ‘chat’ in several areas, including debugging research code (he is a theoretical chemist) and even converting handwritten mathematical notes to LaTeX format.
Hrant does emphasize a nuanced view on the use of AI: “I still write my code from scratch more often than not. What I’ll do occasionally is I’ll have code and I’m not quite sure if maybe it's performing correctly, or there's a bug I've had, AI tools help me track down the source of that bug after I've spent a little time trying to figure it out myself.”
In particular, he has found that AI is a very helpful tool when working with a new programming language — more of an interactive tutor than a solution-granter. “I have a code that's simulating spectral data, and I decided that it would be good to take that data and put it into a script that could then generate a plot of the data to. Directly in GNU plot. And since I'm doing this in Fortran, not something like a Jupiter notebook… it's not a straightforward thing to do - it's not impossible, but ‘chat’ saves me time.”
What else does ‘chat’ help Hrant with?
One of our major topics of discussion, given that Hrant is going to be a key player in developing AI-usage related policy, was the role of AI in academic writing. While some scholars take an adamant stance against the involvement of AI in writing, he takes a much more layered approach, seeing AI as a powerful editing tool that can be useful in finding inconsistencies, suggesting improvements to structure, and even helping to reverse outline a document.
“I've used ‘chat’ to help edit emails and memos and letters and generally, all correspondence that I write,” he says, but “I've told it not to change my sentence structures. I've told it not to delete any sentences, but if it sees a new place, to put a paragraph, to move a paragraph,” he clarifies, highlighting that it is all about keeping one’s own voice and using AI as a means of enhancement. Which explains why Hrant’s emails sound like him, except better!
And what about teaching or advising usage of AI?
“So, the conversation around using AI in our lab is, I would say, evolving. It's come up in multiple contexts. I think we've talked about it in different ways,” he says, and elaborates a bit more about using AI in his graduate class this upcoming semester. “There are good ways to incorporate AI into a high-level graduate course. We do a lot of coding and a lot of formal math in the course... one thought I've had, and I wanted to try this out at some point, is to develop sort of draft mathematical proofs that maybe have errors or holes and have students kind of practice going to the board and working through that.”
Take note, those taking this class!
Challenges and Future Implications
As our discussion turned to broader institutional challenges, Hrant noted the inevitability of artificial intelligence: institutions of learning are faced with a choice only between immediate or delayed adoption, and adoption is inevitable. And while some University of California campuses have taken significant steps in integration of AI in some respects, much work needs to be done across the system given that there is a need for clear directives, rather than complete bans given the growing role of AI in scholars’ future careers.
“The academic community is, in general, slow to change, and much more comfortable with what it knows. And this, this is a disruptive force. It’s only been two years since OpenAI went public, followed by the real sort of second generation learned generative AI tools about six or nine months after that, and then the sort of discussion around how to actually do prompt engineering, well, all that took just a few months.”
What about using AI tools in everyday work, especially by students?
First up, Hrant clarifies “hard boundaries are not yet well defined.” He points out that this is a question of allowing students the use of AI tools while, at the same time, developing critical skills and understanding.
“I think that there has to at least be part of the conversation focused on acknowledging upfront that AI is a reality, and then encouraging everybody, whether students, faculty, or staff, who might participate in any kind of activity around AI, to join in.”
There we have it – a complex balancing act for educational institutions expecting to infuse artificial intelligence into their structures of pedagogy. But at least, UC Merced’s Graduate Dean is on board with AI - he uses ChatGPT for his own work, and thinks universities should embrace it smartly rather than fight it, while emphasizing the need for thoughtful policies that preserve core educational values for the students.
Interested to contribute? Email sghosh@ucmerced.edu
