When Professor G. Sue Kasun of Georgia State University taught a new course this summer, she used artificial intelligence to help her brainstorm.
A professor of language, culture and education, Kasun teaches current and future language educators. She then used Gemini (Google's generative AI chatbot) to come up with ideas for course reading and activities to integrate identity and culture into language education.
“There were suggestions to offer students a variety of options, such as having students generate images and having students write poetry. These are things I can think of, but there are limits to our time.
Kasun uses Gemini to create grading rubrics. She says she constantly checks to make sure what it produces is accurate.
It's a massive time savings, she says.
Kasun uses the AI model generated in work to increase the number of higher education teachers.
A national survey of over 1,800 higher education staff conducted earlier this year by consulting firm Tyton Partners found that around 40% of managers and 30% of instructions use daily or weekly generation AI.
New research from the human race behind AI Chatbot Claude suggests that professors around the world use AI to develop curriculum, design lessons, conduct research, write grants, manage budgets, score student work, and design unique interactive learning tools.
“When I looked up the data late last year, education made up two of the top four use cases in all the ways people use Claude,” says Drew Bent, an education leader in humanity and one of the researchers who led the research.
This includes both students and professors. Bent says these findings have influenced reports on how university students use AI chatbots and reports on the latest research into Claude's use.
How Professors Use AI
Anthropic's report is based on roughly 74,000 conversations users with higher education email addresses had with Claude over 11 days, late May and early June this year. The company used automated tools to analyze the conversation.
The majority (or 57% of the conversations analyzed) related to curriculum development, such as lesson plans and assignment design. One more surprising finding, Bent said, was a professor who used Claude to develop interactive simulations for students, such as web-based games.
“It helps you write code so that you, as an educator, can share with students in your class, and that you can write interactive simulations that will help you understand the concept,” says Bent.
The second most common way that the professor used Claude was for academic research. This consisted of 13% of the conversation. Educators also used AI chatbots to complete administrative tasks, including budget planning, drafting letters of recommendation, and creating meeting agendas.
Their analysis suggests that professors tend to automate more boring and routine tasks, including financial and administrative tasks.
“But in other areas such as education and lesson design, the collaboration process where educators and AI assistants traveled back and forth and worked together was much higher,” Bent said.
The data comes with warnings – humanity has published its findings but did not release the complete data behind it – including the number of professors taking part in the analysis.
The study then took snapshots in time. The period studied covered the end of the tail of the school year. For example, if Bent had analyzed the 11-day period in October, the results could have been different.
Grading student work with AI
Approximately 7% of humanity in the conversations analyzed were related to scoring student work.
“When educators use AI for grading, they often do a lot of automation and do an important part of grading to AI,” says Bent.
The company partnered with Northeastern University on the study. We investigated how and why 22 faculty members use Claude. In their survey responses, university faculty said scoring student work was the job that chatbots were the least effective.
It is not clear whether any of the ratings produced by Claude are the ratings that were actually considered in the grades received by students and feedback.
Nevertheless, Mark Watkins, a lecturer and researcher at the University of Mississippi, fears that human discoveries are showing disturbing trends. Watkins studies the impact of AI on higher education.
“The nightmare scenario of this kind that we may encounter is to use AI to grade the same papers using AI. If so, what is the purpose of education?”
Watkins also says he is wary of using AI, as he says, deciding the relationship between the professor and the student.
“I'm just using this to automate a part of my life, whether it's writing an email to a student or providing a letter of recommendation, rating, or feedback, I'm really against it,” he says.
Professors and faculty need guidance
Kasun, a professor in Georgia, also doesn't think that professors need to use AI for grading.
She hopes there will be more support and guidance on how the university is best to use this new technology.
“We're here, alone in the woods, dodging ourselves,” Kasun says.
Drew Bent says that like humanity, businesses like him should partner with higher education institutions. He notes: “As a tech company, it's not the right way to tell educators what to do or not.”
But educators and AI workers agree that, like Bent, decisions around how to incorporate AI into university and university courses will affect students for the next few years.
