On Jan. 18, ASU became the first university to partner with OpenAI. Lev Gonick, Chief Information Officer of Enterprise Technology, and Anne Jones, Vice Provost for Undergraduate Education, spoke about the current progress and future hopes for the OpenAI partnership.
READ MORE: Charting ASU's future: Navigating OpenAI's first partnership in higher education
Goals for the partnership
According to Gonick, the ultimate goal is to leverage the technology to advance the mission outlined in the ASU charter.
"In fact, that's the answer to almost any question about when we engage with partnerships with other companies," Gonick said. "When we met with the team at OpenAI, there was this rather interesting moment where, describing the ASU charter to them, they were reflecting back and saying this was very much, with a couple of changes, the goals they had for their company."
OpenAI does not utilize ASU’s data to help fine-tune itself. Instead, it benefits from exposure and design resources specific to large universities.
"As a result of us announcing our earliest partnership with OpenAI, they got many, many introductions into the higher education marketplace," Gonick said. "They're also very interested in working with a trusted partner, that’s (ASU), in actually helping to develop the product offering for education at large, especially for higher education."
According to Gonick, ASU currently supports 16 large language models, including ChatGPT, with plans to add support for more in the coming weeks.
"ASU has been in this space well before there was OpenAI, and we've been developing large language models across the University," Gonick said.
Along with technical support, ASU also offers communication support. According to Gonick, while other companies with OpenAI’s reputation might have large media relations teams, OpenAI had just a single person.
"The point there is that they're not big into the overhead of creating public-facing work," Gonick said. "Now, they don't have much of a choice, because they've created all this market demand, there'll be a point … where they’re going to have to take the education market seriously, and not just hype it."
Rollout across the University
Professor support for AI use in their classrooms is currently mixed, with some professors encouraging its use and others banning it outright.
READ MORE: OpenAI and the arts at ASU
According to Jones, who is also a chemistry professor, this is analogous to banning a calculator for certain mathematical operations – a regular practice in STEM classes.
"Now, why would I do that, right? I know, as a professional, I could pull up a spreadsheet or an online calculator or whatever to do calculations whenever I feel like it," Jones said. "But there is value in learning to solve a problem without a calculator. For example, if I wish (for) students to focus on estimating whether a quantity is meaningful, the calculator may get in the way."
While the use of AI will remain in the hands of individual instructors, there are still efforts to encourage its integration into classrooms. In August, the University launched a training course to introduce professors to the use of AI.
"We created, along with EdPlus and the provost office, an eight-module set of learning tools for professors," Gonick said. "From 'I don’t know anything about it,' 'I know enough about it but I’m very skeptical' … all the way up to fairly intermediate uses of generative AI for teaching and learning activities."
According to Gonick, around 1,700 professors have already taken advantage of these courses.
"For reference point, we have about 6,000 faculty," Jones said. "So about 30% of faculty have engaged to some extent in that conversation, which is very high."
Security status
The Family Educational Rights and Privacy Act is a federal law that protects the privacy of students’ personal information. Currently, ChatGPT Enterprise is not approved for FERPA data under ASU’s proposal guidelines, but this is an active project for the partnership.
"The (kind of data) that we're working on with them has to do with obviously PII (personal identifying information)," Gonick said. "The (data) we’re most concerned about at ASU … is making sure we protect student information."
One barrier to assuring a proper level of security is the necessary institution of a failsafe. Even if there is some human mistake or computer error, there needs to be a way to catch it before the sensitive data can spread.
"The code will give us not only a flag, but take that data into a parking lot for human intervention, to make sure that people understand that unless you actually say yes, you’re about to put (PII) information into (the system)," Gonick said.
Currently, this feature is under development, but students are expected to have access to the updated systems as early as next semester.
"I think students will see services that will have all the right protections and guardrails by the time (students) come back in August," Gonick said. "That’s our timeframe for working with OpenAI on that."
Proposal project development
The AI Innovation Challenge opened to students on March 25, allowing them to propose a project using ChatGPT Enterprise, given that they have an advising professor.
"You also have access to our team," Gonick said. "We have a 17-member AI acceleration team made up of data scientists and(machine learning) specialists and deep learning specialists who are essentially consulting services to the grant recipients."
Evaluation of student proposals will be done by the same panel that looks at faculty proposals. This includes reviews by the Office of the University Provost, Enterprise Technology, EdPlus, faculty members of the ethics committee and a team for regulation and compliance.
"All of this is in the abundance of caution mode," Gonick said.
Looking ahead
When asked about the future years of AI, both Gonick and Jones agreed that the exploration process is only just beginning.
"As much excitement as there is now, it’s more like just the opening credits of a film,” Gonick said. "In the next five years, I really think we’re going to have a mature offering for personalized or customized tutors … or study buddies."
Jones instead compared AI’s future in higher education to the introduction of the internet, ideally improving quick access to knowledge if one knows how to search properly.
"I think AI has the potential to make education more human," Jones said. "Which sounds perverse, but if you think explicitly about how AI (takes) away some of the (routine) aspects of accessing knowledge, that then leaves space for humans to have conversations that are more human."
Edited by Shane Brennan, Sadie Buggle and Angelina Steel.
Reach the reporter at asgrazia@asu.edu and follow @emphasisonno on Twitter.
Like The State Press on Facebook and follow @statepress on Twitter.
River is a senior studying english and mechanical engineering. This is their fourth semester with The State Press.
