AI developer Advait Paliwal released Einstein AI earlier this year, a software he claimed could log into Canvas, do your homework and automatically submit it for you.
In February, the software was taken down following a cease and desist for a trademark infringement, but Tulane University professors say this is not the last we will see of autonomous AI in higher education.
Canvas is a learning management system used at Tulane and other universities across the nation. Unlike traditional AI, which responds to human prompts, agentic AI systems such as Einstein use human-like reasoning to act independently without constant input. Einstein is not currently available for download.
“My first question is, is it even real?” Nicholas Mattei, associate professor of computer science, said. “Maybe not. Then there’s the thing that [Paliwal] claims the software can do, right? Be a student inside Canvas … That’s very real … If it doesn’t exist now, it will in six months.”
According to Ralph Russo, director of the Tulane School of Professional Advancement Information Technology Programs, agentic AI programs could pose security risks for users.
“If this software is going to mimic you as a student, then you’re going to have to give it your credentials so it can log in as you work … From a cyber security perspective, giving anybody or anything your credentials is a serious flaw,” Russo said.
However, Russo said that Tulane can detect when students are using agentic AI by reviewing the IP addresses that access the system.
According to Russo, Tulane can preemptively block these IP addresses, but AI companies can then change them, contributing to what Russo described as an “arms race.”
Daniella Zalcman, Tulane communications professor, expressed concern over how agentic AI will harm students’ ability to think.
“Creating a piece of software sophisticated enough to essentially act on your behalf is just going to completely erode the idea of what it means to get a college education,” Zalcman said.
As AI technology progresses, professors are changing their class structures. Mattei moved much of his classes’ work to on paper and now focuses on students’ ability to explain concepts rather than complete them.
“On the teaching side, how do we do things in the classroom that make sure these things are valuable?” Mattei said.
Despite concerns, professors still admit positives in the present and future of AI.
“This is growing pains for AI … It’s going to be great for students, but there are dangers, and there is misuse … [you] might think they are cheating the university, but you’re really cheating yourself,” Russo said.