New ChatGPT Study Mode walks students through problems instead of solving them

OpenAI launched a new “Study Mode” for ChatGPT on July 29, designed to guide students through learning processes rather than providing direct answers, as educators grapple with rising AI misuse in academic settings.

The feature uses Socratic questioning techniques to guide students through problems step-by-step. Study Mode is available immediately for ChatGPT Free, Plus, Pro, and Team users, with ChatGPT Edu access coming in the next few weeks.

“When ChatGPT is prompted to teach or tutor, it can significantly improve academic performance,” said Leah Belsky, OpenAI’s VP of Education. “But when it’s just used as an answer machine, it can hinder learning.”

A Guardian survey found almost 7,000 proven cases of AI cheating in UK universities during 2023-24, compared to 1.6 cases per 1,000 students in 2022-23.

More than a third of college-age Americans use ChatGPT, according to OpenAI. About a quarter of their messages involve learning, tutoring, and schoolwork.

Study Mode combines interactive prompts, hints, and self-reflection questions to guide understanding. The system organizes information into easy-to-follow sections and tailors lessons based on skill level assessments from previous conversations.

OpenAI developed the feature with help from teachers, scientists, and education experts. The company tested it with college students who provided feedback during development.

“The best way I’d describe it is a live, 24/7, all-knowing office hours,” said Noah Campbell, a college student who tested the feature.

However, the tool has significant limitations. Students can easily toggle Study Mode off to get direct answers whenever they want. OpenAI currently offers no parental or administrative controls to lock students into the learning mode.

The company acknowledges that Study Mode may show “inconsistent behavior and mistakes across conversations” since it runs on custom system instructions rather than a new AI model.

OpenAI plans to add clearer visualizations, goal tracking, and deeper personalization features. The company is also partnering with Stanford’s SCALE Initiative to study how AI tools affect learning outcomes in K-12 education.

Anthropic launched a similar “Learning Mode” for its Claude chatbot in April. Both companies are racing to capture the lucrative education market while addressing widespread concerns about AI enabling academic dishonesty.

The ultimate test will be whether students choose to engage with the guided learning experience or simply switch back to getting quick answers when looming homework deadlines approach.