OpenAI Seeks To Combine Chatbots And Online Classes

The next online course you take may include a chatbot if OpenAI has its way.

Siya Raj Purohit, a member of OpenAI’s go-to-market team for education, stated during a fireside on Monday organized by Coeus Collective that OpenAI may look into ways to allow e-learning teachers to construct unique “GPTs” that connect to online curricula.

“I’m hoping that instructors will make personalized GPTs for the general public and allow people to interact with the material in a way that lasts a lifetime,” Purohit stated. “It’s on the roadmap, but it’s not part of the work we’re doing right now.”

According to Purohit, she has already seen instructors use OpenAI’s built-in capabilities to generate unique GPTs by uploading a “semester’s worth” of content, which they then make accessible to their students. Students interact with that limited knowledge, she continued, “[which] is a really powerful and good way to let them research.”

Leah Belsky, the former chief revenue officer of Coursera, was appointed by the firm in September as its first general manager for education, with the responsibility of expanding the use of OpenAI’s products in educational institutions. Additionally, ChatGPT Edu, a version of ChatGPT designed for academic institutions, was introduced by OpenAI this spring.

Within the next ten years, the market for AI in education could be valued at $88.2 billion, according to Allied Market Research. However, skeptic educators are mostly to blame for the slow start to growth.

The GPTs Purohit outlined may resemble Khanmigo, a chatbot developed by Khan Academy, an online learning platform that was introduced last year in partnership with OpenAI. By closely integrating with Khan Academy’s educational resource repository, Khanmigo can assist students with homework assignments, test preparation, and other tasks.

Khanmingo’s blunders highlight the current drawbacks of AI. The chatbot had trouble understanding simple math when tested by The Wall Street Journal in February, and it frequently failed to fix mistakes when asked to double-check answers.

But Purohit said the technology is becoming better.

“We want to help translate that into what works in learning and teaching, and all of our models keep getting better,” she stated.

Most educators are still dubious. According to a Pew Research Center survey this year, 25% of public K–12 educators believe that implementing AI tools in the classroom causes more harm than good. According to a different survey conducted by the Center for Reinventing Public Education and the Rand Corporation, only 18% of K–12 teachers are using AI in the classroom.