Inside Khan Academy’s Collaboration with OpenAI: Bringing AI into Education

Sal Khan, founder of Khan Academy, initially declined OpenAI’s offer to integrate ChatGPT technology at his educational platform. Despite OpenAI’s research lab being relatively unknown, and ChatGPT-3 not functioning as expected, Khan’s refusal highlighted his commitment to maintaining integrity.

In 2008, Khan Academy was established as a nonprofit aiming to provide free, high-quality education globally. With a vast library of video lessons and interactive exercises, its services are utilized by approximately 190 million people worldwide, and nearly 800 school districts in the United States. This success has made Khan a reputable figure in education, attracting many organizations seeking to collaborate.

When OpenAI approached Khan for the second time in summer 2022, the conversation included co-founder Sam Altman. Prior to ChatGPT-3.5’s launch, Khan and his chief learning officer, Kristen DiCerbo, signed a nondisclosure agreement. They participated in a Zoom call revealing GPT-4, showcasing advanced capabilities far superior to previous models.

Khan immediately saw the potential GPT-4 held for education. The model could not only correctly answer Advanced Placement Biology questions but also explain its reasoning for correct and incorrect answers. Additionally, it could generate new exam questions with endless possibilities, demonstrating a significant leap in generative artificial intelligence.

Recognizing the impact of these advancements, Khan was initially conflicted about collaborating with OpenAI. Although GPT-4 was not perfect, the technology represented a major advancement in AI. OpenAI aimed to launch GPT-4 shortly, encouraging Khan Academy to develop a tutoring bot for student use concurrently.