Khan Academy and OpenAI’s Classroom Innovation

In early 2021, Sal Khan, the founder of Khan Academy, initially declined an invitation from OpenAI’s Greg Brockman to explore the potential of ChatGPT for educational purposes. At that time, OpenAI was not widely known, and ChatGPT-3 was akin to an early experimental model. It often displayed bursts of intelligence but lacked consistency.

Khan Academy has been a respected figure in education since its founding as a nonprofit in 2008, providing a vast array of free educational resources. The platform serves around 190 million users globally and is used in nearly 800 school districts in the U.S., reinforcing its commitment to accessible, high-quality education.

OpenAI sought out Khan for his esteemed reputation across political and educational spectrums. The first outreach did not lead to a partnership, but a subsequent one, involving OpenAI’s Brockman and Sam Altman, presented a more compelling case. In the summer of 2022, Khan and his chief learning officer, Kristen DiCerbo, learned about GPT-4, OpenAI’s upcoming generative AI model, under a non-disclosure agreement.

The potential of GPT-4 quickly impressed Khan. The model wasn’t only proficient in answering complex questions, but it also explained its reasoning and could create new, varied exam questions. This capability signified a major educational breakthrough, despite initial imperfections.

Khan was conflicted about integrating OpenAI tech into his life’s work at Khan Academy. The weight of this decision lay in the significant impact it could have on education.

With GPT-4 set to launch within six months, OpenAI aimed for Khan Academy to introduce an AI-powered tutoring bot simultaneously. This bot would allow students to access enhanced learning directly from their laptops, marking a significant advancement in AI application within education.