Skip to content

OpenAI’s New ChatGPT Edu for Universities: Will Students Benefit?

In May 2024, OpenAI announced ChatGPT Edu, a new ChatGPT version designed for colleges and universities that would like to provide artificial intelligence capabilities for their students, faculty, staff, and researchers. Following the company’s early 2024 deal with Arizona State University, the new platform amounts to another one of the company’s first forays into the higher education marketplace.

At first, it wasn’t clear why a partnership between an R1 research university like Arizona State and an artificial intelligence software developer might have emerged, and we first speculated on those reasons in our April 2024 feature article entitled “Arizona State and OpenAI Announce New Partnership—But Why?” Now, four months later in July, the factors driving such an initiative seem more apparent.

This new Edu platform is strategically important for OpenAI as the company attempts to overcome resistance to artificial intelligence among college faculty. With the possible exceptions of departments such as computer science and electrical engineering, higher education faculty tend to adopt new innovations slowly.

But generative artificial intelligence has attracted an intense level of resistance among many academics, as we point out in our January 2024 feature article, “Are College Faculty Boycotting Artificial Intelligence Tools Like ChatGPT?” That report describes a surprising poll by Tyton Partners showing that by fall 2023, only about 20 percent of United States college faculty had tried AI tools more than twice.

Why was that result so controversial? About 75 percent of the faculty AI users—and even half of the faculty nonusers—all told the pollsters that their students won’t succeed in the labor force without understanding how to use artificial intelligence tools.

In other words, even though most faculty believed that skills in operating these AI technologies are crucial for their students’ success in the labor market, the overwhelming majority of those instructors still refused to use any AI tools as late as 11 months after ChatGPT’s launch on November 30, 2022. Many of the new capabilities in the Edu platform could help overcome that widespread faculty resistance by addressing several of their concerns.

ChatGPT Edu’s Extra Capabilities

This new version appears to offer advantages not available to universities in other OpenAI products or only available at higher prices. For one thing, ChatGPT Edu offers universities much higher message limits than those available with Chat GPT’s free version. That should make the platform better able to perform various tasks across campus that OpenAI promotes, including personalized student tutoring, grant proposal preparation by researchers, and grading by faculty.

The Edu package also offers substantial advantages in security and privacy. For example, ChatGPT Edu offers “walled garden” security and privacy controls similar to those within ChatGPT Enterprise, the version originally delivered to Arizona State. Several ASU faculty members reportedly objected to that deal until OpenAI agreed to provide the university with the privacy and security capabilities sold with the Enterprise version. The enhanced privacy features also mean that the Edu package will not train future versions of the GPT large language models by using data shared by university users; their data will be considered confidential and proprietary by the system.

The Edu version is also powered by GPT-4o, OpenAI’s latest and most advanced large language model, which the firm demonstrated live via YouTube in May 2024. The firm claims this new language model can “reason across text and vision and use advanced tools such as data analysis.” There’s no question that college faculty, researchers, and students would certainly find new capabilities like those attractive.

However, it remains to be seen which of GPT-4o’s new capabilities OpenAI will actually provide to universities and which will turn into illusory “vaporware” functionality the firm promised but never delivered. Brian Chen of the New York Times even goes so far as to argue that the May demo turned out to be a hyped “bait and switch.” That’s because the company subsequently released GPT-4o with many of the demonstrated features missing.

For example, the demo’s new voice with so many human-like inflections and emotions was one of those deleted features. It was apparently removed after lawyers for actress Scarlett Johansson asked OpenAI to reveal how the firm developed the demo’s voice; she claimed the new voice sounded uncannily similar to her own.

Johansson had performed as the voiceover talent for the speech of Samantha, an artificially intelligent virtual assistant in the provocative 2013 film Her, and a persona with whom Theodore—the main character going through a bitter divorce—falls in love. Johansson now claims that OpenAI’s CEO Sam Altman had personally asked her twice to provide a new speaking voice for GPT-4o, but she refused both offers.

After Johansson threatened litigation against OpenAI, the company replaced the demo’s voice known as Sky with a different one that lacked all the improvements. And even though the chatbot provides faster and more lucid speech replies—and now can even translate spoken language in real time—at the time of this writing in July 2024, the version available to the public still responds in one of the older, more robotic-sounding voices.

Moreover, Chen also criticized a segment of the online demo that showed how the new chatbot tutored a high school student in solving a geometry problem. EdSurge’s Jeffrey Young describes the demo this way:

Sal Khan, founder of Khan Academy, which has been building a tutoring tool with ChatGPT, sits watching his 15-year old son Imran learn a math concept from a talking version of the chatbot running on an iPad, which can also see what the student is typing on the tablet.

As Sal Khan looks on nodding, the chatbot asks his son a question in a friendly female voice about triangles, and Imran answers while indicating which side of the triangle he means using a stylus and tapping on the iPad screen. It’s an interaction that might have seemed like science fiction a couple of years ago.

Unfortunately, several aspects from this video’s demo are misleading. First, Chen points out that GPT-4o’s ability to provide real-time analysis of such a math problem by using the phone’s camera depicts another feature that isn’t available. According to Khan, that technology is at least six months to a year away.

Second, Khan’s son Imran doesn’t actually learn a new geometry concept during that demo, as the video implies. Instead, this gifted student actually took calculus while he was in the seventh grade, which means he probably had completed plane geometry during roughly the fifth grade. Because Imran would have learned the definition of a triangle’s hypotenuse about five years ago, that means he and his father are merely playing roles as high-profile celebrity actors during the demo.

ChatGPT’s Success at Universities

In a prepared statement, OpenAI says it built ChatGPT Edu because it had observed certain universities’ success with the Enterprise version. The firm cited schools like the Wharton School of the University of Pennsylvania and Arizona State among the use cases.

Here at OnlineEducation.com, we’ve covered work using ChatGPT by the University of Pennsylvania’s Professor Ethan Mollick in his MBA classes at the Wharton School since January 2023 and again the following April. Because he was one of the earliest ChatGPT pioneers in higher education, seeing OpenAI cite one of Dr. Mollick’s assignments wasn’t surprising.

In the cited use case, MBA students and undergraduates in Dr. Mollick’s courses finished their final reflection assignments by chatting with a GPT specifically trained on materials in their course. OpenAI reported that ChatGPT “got them to think more deeply about what they’ve learned.”

We also weren’t surprised to see a GPT-based tutoring system developed by Arizona State for its German language students listed as another one of the use cases. That’s because in the conclusion of our article on the university’s partnership with OpenAI, we covered ASU President Michael Crow’s statements that tutoring app development was a key focus within the research enabled by the university’s OpenAI deal.

Automated User Provisioning

Along with the enhanced security, the Edu version also includes hidden advances that speed up access to ChatGPT accounts for new users. One of these is known as automated user provisioning, or AUP, and it’s a huge time- and labor-saving innovation for system administrators at universities.

In short, provisioning is the process of setting up new user accounts with the correct permissions. Although this procedure can be done manually by the technical specialists at large corporate enterprises that hire infrequently, it’s not practical for a university with thousands of new student users every semester, or—as with many of the larger, predominantly online colleges—every several weeks throughout the year. Arizona State’s Program Manager Paul Alvarado describes the process this way:

One of the biggest challenges was managing the scaling at our level—such as provisioning and deprovisioning user access at scale. With students and faculty members frequently entering and leaving our system, manually managing permissions is not feasible.

We worked with OpenAI to automate this process effectively, ensuring our system integrates seamlessly with theirs. This not only helps us but also benefits countless other institutions of higher learning.

Is ChatGPT Edu More Affordable?

ZDNet’s Editor Sabrina Ortiz says that the biggest selling point for the Edu package is that it’s more affordable than ChatGPT Enterprise. The lower price point would presumably benefit students at more universities across the nation because it might lead colleges who balked at paying premium prices for the Enterprise version to purchase the Edu configuration. But is she correct?

Like most companies that sell to large organizations in the enterprise and university markets, OpenAI does not comment on its pricing. However, that policy hasn’t stopped unconfirmed reports from surfacing in the media.

Citing one such unconfirmed source from September 2023 on Reddit, a June 2024 TechCrunch article reports that OpenAI’s monthly price for the Enterprise package sells for about $60.00 per user, with minimums of 150 users and a one-year contract, which works out to a minimal charge per organization of about $108,000 annually. Plus an unconfirmed reader comment below the same article also reports that the monthly price for the Edu package sells for about $12.00 per user, which is only about 20 percent of the Enterprise version’s price.

However, that reader comment also claims the minimum number of Edu users is set much higher to about 10,000. That would work out to a minimal site fee of about $1.44 million annually.

If true, that pricing might help America’s largest colleges like Southern New Hampshire University or Western Governors University save money over purchasing the Enterprise configuration. However, that constraint would not help smaller colleges and universities with fewer than 10,000 students who—unless they negotiate a special sales order—presumably would still need to purchase the Enterprise version anyway.

Douglas Mark

While a partner in a San Francisco marketing and design firm, for over 20 years Douglas Mark wrote online and print content for the world’s biggest brands, including United Airlines, Union Bank, Ziff Davis, Sebastiani and AT&T.

Since his first magazine article appeared in MacUser in 1995, he’s also written on finance and graduate business education in addition to mobile online devices, apps, and technology. He graduated in the top 1 percent of his class with a business administration degree from the University of Illinois and studied computer science at Stanford University.