Stanford’s ‘QuizBot’ helps students retain 25 percent more information

Researchers at Stanford University have developed a new chatbot — personified as a penguin named Frosty — to improve the way that students study outside the classroom by making learning more engaging and personalized.

“A significant amount of learning involves factual knowledge and sometimes [knowledge] needs to be learned outside of a traditional classroom setting,” said Sherry Ruan, a doctoral computer science student at Stanford University. “However, current electronic tools, such as flashcards … involve passive learning, and as a result, their effectiveness really relies on people wanting to use them.”

The chatbot, which its creators have named QuizBot, is different.

Quizbot uses algorithms for natural language processing, which Ruan told EdScoop are becoming more sophisticated. The technology allows users to speak or type out their answers during a two-way dialogue and receive feedback from a bot, creating a more engaging human-like experience.

“We want to provide better-automated ways to help students learn factual knowledge in a more fun and effective way,” she said.

In a study of 36 students, Stanford researchers found that students who had been provided Quizbot spent 2.6 times longer studying than those who had been provided a flashcard app. Students who used Quizbot also correctly recalled correct answers 25 percent more frequently.

“To give QuizBot a more human touch, the chatbot is personified as an animated penguin named Frosty, who interacts with the user in various ways,” Ruan said.

A typical round of conversation begins with Frosty asking the user a factual question. Frosty may ask a biology student, “What is the diffusion of water through a semipermeable membrane down its concentration gradient?” The user can then responds by typing out an answer. However, one hurdle in designing such a tool is that QuizBot must be able to recognize correct answers in various forms. Students may use various words, syntax and spelling. That’s where artificial intelligence comes in.

An algorithm processes the user’s answer and evaluates it against the correct answer. If right, Frosty congratulates the user with: “That’s correct!” Otherwise, Frosty acknowledges that the user answered incorrectly before displaying the correct answer. If a student is stuck on a question, Frosty will offer several answer choices to choose from before moving on.

“One drawback is that it often takes more time to practice one question with QuizBot than it does to go through one flashcard,” Ruan said. “Chatbots are likely not a good means of cramming before an exam.”

However, according to the research, longer study time is a key part of why QuizBot works so much better.

Although chatbots are widely used in industries like customer service and banking, they are severely underutilized in the field of education, Ruan said.

“QuizBot demonstrates that chatbots not only have a place in the future of education, but they also have the potential to surpass the efficacy of traditional methods while boosting engagement in learning,” she said. “From a learning perspective, this is exactly what we want out of the future of educational technology.”

QuizBot is a part of a larger AI Education research project at Stanford called Smart Primer, which is aimed at developing an adaptive tutoring system using narrative, chatbots and real-world activities.