Case Study: Streamlining GenAI education in the university classroom
How Carnegie Mellon University uses BricksAI to easily grant student access to GenAI tools.
Introduction
User: Carnegie Mellon University’s METALS program
Industry: Higher education
As AI literacy has become an increasingly important skill, many universities have recognized the urgent need to incorporate generative AI technologies into their curriculum. Carnegie Mellon University’s Master of Educational Technology and Applied Learning Science (METALS) program is no exception.
The METALS program is a one-year, interdisciplinary master’s program that trains graduate students to become learning engineers and learning experience designers. The program culminates in a seven-month capstone project, in which teams of students work with an external client to solve a real-world education problem.
The Challenge: A Secure Way to Share LLM Access with Students
With the rapid advancement of generative AI technologies, many students expressed interest in using services like OpenAI to build their solutions. However, when the METALS program began incorporating OpenAI into their curriculum, they immediately faced several issues:
Firstly, there was no straightforward way to grant students access to OpenAI. The METALS program was uncomfortable with giving students OpenAI API keys to the school’s account, as leaked keys could allow bad actors to quickly drain the organization’s budget and rate limit.
Ultimately, each student team was asked to create an OpenAI account, pay for the usage upfront, and then go through a reimbursement process with the school. This process created a substantial administrative burden for the teaching staff.
“Prior to BricksAI, we had to manually process receipts from each team post-project — a process that was both inefficient and caused delays. BricksAI completely transformed our approach.”
Gautam Yadav, Senior Learning Engineer @ Carnegie Mellon University
Secondly, new accounts created by students are subject to OpenAI’s Free Tier limits, which means students do not have access to the latest GPT-4 models and are subject to strict rate limits. To work around this issue, students have to pay $5 upfront to upgrade to Tier 1 to gain access to GPT-4, or $50 for Tier 2 if they need a higher rate limit, even if they do not use all the credits.
Solution: Using BricksAI as a centralized platform for managing OpenAI access
To address these challenges, the METALS program adopted BricksAI as a centralized platform for managing OpenAI access.
With BricksAI, instead of requiring each student team to create an individual OpenAI account, the METALS program could create one account for the entire class. Then, using the BricksAI dashboard, the teaching assistant could allocate a fixed budget and rate limit to each team.
Additionally, BricksAI provides real-time monitoring of all OpenAI usage. Teachers can easily adjust the limits according to each team’s needs. They can also review logs of each student’s LLM usage in order to provide help.
Impact: Transforming GenAI in classroom
This transformation, powered by BricksAI, enabled students and teachers to dive into the world of generative AI without worrying about administrative work.
Looking ahead, Yadav reveals, “We’re excited about extending BricksAI’s benefits to other technical courses within our curriculum.”
The CMU METALS program’s journey with BricksAI has shown the immense potential of the platform not only in enterprise settings, but also in education.
At BricksAI, we’re committed to empowering educational institutions and learners with the tools needed to navigate the rapidly evolving landscape of generative AI technologies. By simplifying access to advanced AI resources, we aim to democratize the learning process, ensuring that students from all backgrounds have the opportunity to engage with cutting-edge technologies.
