Description of the video:
Hi everyone, my
name is Aarav Kumar and today I would
like to show you our IU Undergraduate
Research GPT. Many undergraduate
students at Indiana University struggle
to find research opportunities because
information about labs, faculty, research
areas are scattered across different
department websites. Often, these students
feel overwhelmed and they're not really
sure where to start and they're kind of
intimidated by the process of contacting
these professors. Our project, the IU
Undergraduate Research GPT, was designed
to simplify this process by helping
students discover research opportunities and
narrow down their interest before moving
on to an outreach in a more guided and
personalized way. So our custom GPT uses
structured prompting, department level
matching, and official IU source links in order
to help the students navigate these
research opportunities. Our GPT is
organized into four
different phases. Get to know you phase,
we have the exploration and department
matching, then third we have a lab and outreach
support, and lastly we have a final report
or a restart option. In the get to know
you phase here, the GPT begins by
explaining the process and asking questions
one at a time to better understand the
students' interests, major skills,
and experience. For example, I'll
say I'm a sophomore at IU studying
neuroscience. I'm interested in
Alzheimer's research, but I don't have any
research experience. Instead of
immediately giving random results, the
GPT builds context before making
recommendations. Next, the GPT moves into
the exploration phase using department
and faculty data. It suggests research
areas and departments connected to the
student's interest. It can also suggest
interdisciplinary paths students may not have
already considered. For example, the
GPT narrows results towards Alzheimer's
and memory cognition research and
identifies relevant IU Bloomington departments
and faculty. After students
identify a lab they're interested in, the GPT
helps them explore the opportunity further
and take the next step for example here
the student selected the comparative
cognition laboratory led by dr jonathan crystal
which focuses on memory cognition and
related neuroscience research the gpt
provides additional information about the
lab and then helps students prepare
professional outreach communication in this
example the student asked the gpt for help
writing an email asking about research
opportunities in the lab. We intentionally designed this feature to
guide students through the
outreach process without completely
replacing authentic student
communication. Throughout
development, we encountered
several challenges. One major challenge
was that university lab information was
often incomplete or constantly changing. To
address this, we used structured department
links and refined prompts to improve
accuracy and consistency. We also discovered
through testing that users interacted
more naturally with conversational guidance
rather than rigid instructions, so
we redesigned the GPT to allow more
flexible interactions. Overall, the IU
undergrad research GPT was designed
to make research opportunities more
accessible and less intimidating
for students. Rather than replacing
student effort, the tool supports
exploration, organization, and
outreach by guiding students through a
process that many undergraduates currently
navigate alone. While the project is
still evolving, it demonstrates how
generative AI can be used to create practical,
student-centered support systems within
higher education.
