Whether on college campuses or in middle school classrooms, the advent of artificial intelligence has sent shock waves throughout the academic world. Educators are beginning to reckon with the new reality that AI tools are readily available to students. To tackle this issue, one UC Berkeley student is working on a compromise between banning AI in schools entirely and letting students use AI freely.
In 2023, Haas School of Business student Ghazaleh Sadooghi and two co-founders created Rumi, an educational tool for schools and universities. Using Rumi, students write on a document where depending on what level of AI help their teacher allows for that particular assignment, they can get grammar edits and feedback from AI.
Rumi also has a built-in AI text generator, connected to Gemini and the latest version of ChatGPT. Students can copy and paste text from Rumi’s text generator, but if text is pasted from an external source, it is marked as such.
Once an assignment is submitted to the instructor, Rumi assigns it a rating out of five stars, based on metrics such as number of edits and time taken to write, with more stars meaning more originality.
If the instructor wants more detail, they can examine the document’s editing history from start to finish, including copy and pasted paragraphs from internal and external sources. If text has been pasted from Rumi’s internal tool, instructors can see what prompts students used.
Since its creation, the Rumi team has grown to five people as the service expands its client base. Currently, Rumi is being used at Haas, Tufts University, Foothill College and several K-12 schools in Pennsylvania.
Sadooghi was taking MBA classes at Haas when ChatGPT was released. A professor whose exams were online asked Sadooghi and her classmates to take the exam using a proctoring tool. They had to sit in front of a camera, not moving out of frame or looking anywhere but the screen.
“I thought, ‘This can’t be the new reality,’ ” Sadooghi said. “As a student, I don’t (want) this to be the new reality of my life, not because I want to cheat, but because I don’t want to be video recorded and stare at the camera while I’m taking an exam.”
She talked to her co-founder, Mo Zadeh, who had experience developing Google’s text platform Google Docs. For Zadeh, it immediately clicked that if instructors could see a document’s editing history, they could easily verify who had done the work.
Sadooghi explained that before Rumi, there were two categories of people — those who wanted AI to be banned in schools entirely, and those who thought that since AI was being used in the workplace, students should be able to freely use it too.
“We think AI is good. But we also understand the other side of the spectrum, that we want to allow the students to think critically and build and generate their own responses and be authentic rather than just be able to ask the question and get whatever AI is sending out,” Sadooghi said. “It’s the balance and the art of balancing in each situation what works best.”
