Researchers from Stanford University developed an AI system calledTutor CoPilot on top of OpenAI’s GPT-4 and integrated it right into a platform called FEV Tutor, which connects students with tutors virtually. Tutors and students type messages to at least one one other through a chat interface, and a tutor who needs help explaining how and why a student went flawed can press a button to generate suggestions from Tutor CoPilot.
The researchers created the model by training GPT-4 on a database of 700 real tutoring sessions during which experienced teachers worked on on one with first- to fifth-grade students on math lessons, identifying the scholars’ errors after which working with them to correct the errors in such a way that they learned to grasp the broader concepts being taught. From this, the model generates responses that tutors can customize to assist their online students.
“I’m really excited in regards to the way forward for human-AI collaboration systems,” says Rose Wang, a PhD student at Stanford University who worked on the project, which was published on arXiv and has not yet been peer-reviewed “I believe this technology is a big enabler, but provided that it’s designed well.”
The tool isn’t designed to truly teach the scholars math—as an alternative, it offers tutors helpful advice on learn how to nudge students toward correct answers while encouraging deeper learning.
For instance, it will probably suggest that the tutor ask how the scholar got here up with a solution, or propose questions that might point to a unique solution to solve an issue.
