ChatGPT has just launched a new feature for learners—Study Mode on ChatGPT—a learning experience that helps learners solve problems step by step instead of just receiving the answer. This feature will be rolled out to users of the Free, Plus, Pro, and Team plans, and will soon be available on ChatGPT Edu in the coming weeks.
ChatGPT is becoming one of the most widely used learning tools in the world. Students turn to this tool to study through challenging homework, prepare for exams, and explore new concepts. However, the use of this tool in education also raises an important question: how can we ensure it is used to support genuine learning, rather than simply providing solutions without helping students truly understand the material?
ChatGPT has developed Study Mode to help answer this question. When using Study Mode, students are guided by leading questions that tailor responses to their goals and skill levels, helping them develop deeper understanding. Study Mode is designed to be engaging and interactive, aiming to help students truly grasp the material rather than just complete tasks.
How was this Study Mode built?
Fundamentally, Study Mode is based on a foundation of custom system instructions designed in collaboration with teachers, scientists, and educational experts to simulate a set of core behaviors that support deeper learning, including: encouraging active participation, managing cognitive load, proactively developing metacognition and self-reflection, nurturing curiosity, and providing realistic and constructive feedback. These behaviors are all grounded in long-term learning science research and help shape how Study Mode responds to students.
“Instead of doing the work for students, Study Mode encourages them to think critically about the content they are learning. Features like this are a positive step toward effective use of AI in education. Even in the AI era, learning will be most effective when students are engaged and actively participate in lesson content.” — Robbie Torney, Senior Director of AI Programs at Common Sense Media.
Main Features
Interactive prompts: Combines Socratic questioning, suggestions, and self-reflection prompts to guide learners to explore and think actively instead of providing direct answers.
Scaffolded responses: Information is organized into easy-to-follow sections, highlighting key connections between topics, keeping information engaging with just enough context, and reducing overwhelm for complex subjects.
Personalized support: Lessons are tailored to the user's appropriate level, based on skill assessment questions and memory from previous chat segments.
Knowledge checks: Quizzes and open-ended questions, along with personalized feedback to track progress, support knowledge retention, and the ability to apply that knowledge to new situations.
Flexibility: Easily toggle Study Mode on and off during conversations, allowing learners to flexibly adjust their learning goals in each chat.
How students use Study Mode
Study Mode is built to suit university students. Here are some student opinions we received during the initial testing phase:
“To me, the best way to describe it is as a direct ‘office hour’ that’s available 24/7 and has comprehensive understanding.” — Feedback from Noah Campbell, university student
“Study Mode did a great job breaking down difficult materials into clear, coherent explanations.” — Shared by Caleb Masi, university student
“I tried Study Mode to get guidance on a concept I had tried to learn many times before: sinusoidal positional encoding. This mode is like a tutor who never gets annoyed by my questions. After 3 hours of work, I finally understood it well enough to feel confident.” — Maggie Wang, university student
Ways of using Study Mode inspired by university students in the ChatGPT lab
User: I want to learn about Game Theory, specifically the broad scope this field covers, and also the ways you think it could be useful to me in everyday life. Of course, I want you to indulge my curiosity, but mainly you’ll guide me through it and lay out a detailed plan so I can cover the whole scope here. I’ll ask when I’m curious, but definitely guide me quickly through this.
Chat GPT: Perfect. Here’s the general plan I’ll guide us through—think of it as a multi-layered roadmap from foundational ideas to practical intuition:
🔹 Stage 1: Core Foundations
1. What is Game Theory really?
Strategic interaction, rational agents, assumptions.
2. Types of games
Cooperative and non-cooperative
Simultaneous vs. sequential
Zero-sum vs. non-zero-sum
3. Basic concepts
Players, strategies, payoffs, information
Dominant strategy
Nash equilibrium (the core of almost everything)
🔸 Stage 2: Classic Games and Concepts
4. Famous games
Prisoner’s dilemma
Chicken
Battle of the sexes
Stag hunt
Matching pennies
5. Mixed strategy equilibrium
When no pure strategy works
Randomization and expected payoffs
🔹 Stage 3: Deepening the Model
6. Repeated games & Folk theorem
Cooperation emerging over time
Tit-for-tat, Grim Trigger, etc.
7. Extensive form games
Game trees, backward induction
Subgame perfect equilibrium
8. Bayesian games
Games with incomplete information
Beliefs, types, and expected utility
🔸 Stage 4: Real-life Relevance
9. Mechanism design / Reverse game theory
Voting systems, auctions, contracts
10. Behavioral game theory
Bounded rationality, fairness, search methods
11. Applications in everyday life
Negotiation, dating, politics, AI systems, etc.
🔹 Stage 5: Advanced & Optional
12. Evolutionary game theory
Strategies reproducing over time
13. Algorithmic game theory
Network effects, internet auctions
14. Your world: How it matters for Parker
System design, incentives, power structures
Throughout this process, I’ll incorporate:
🔄 Guided examples
🧠 Thought experiments
🎭 Real-world scenarios to apply the theory
🧩 Quick checks to ensure ideas are landing
Let’s start Stage 1.
First quick question: In your own words, what do you think Game Theory is? No need to polish—just say it.
User: I think Game Theory is a kind of metaphysical framework for aspects of modern society in life, where we consider people in complex situations and systems and try to analyze how people in them make decisions, sometimes why they make those decisions, and how this affects the situation/system.
Chat GPT: Excellent—you’ve circled right around the core.
Let’s dig a little deeper together. Game Theory is basically about strategic decision-making—situations where each person’s outcome depends not only on what they do, but also on what others do.
Quick question:
If you’re making a decision and that decision is entirely dependent on nature or chance—like rolling dice—is that Game Theory?
And then:
If you make a decision and that decision depends on the actions of others—such as whether they cooperate or betray you—is that Game Theory?
Share your thoughts on both.
................
Limitations and Next Steps
This is the first step in a long journey to improve learning on ChatGPT. Currently, Study Mode is based on custom system instructions. ChatGPT chose this approach because it allows for rapid learning from real user feedback and improves the experience—even if it results in some inconsistent behaviors and errors in conversations. ChatGPT plans to directly train this behavior into its core models after finding the most effective method through iteration and feedback from learners.
We are also exploring new features to make Study Mode more engaging and useful for learners, including:
- Clearer illustrations for complex or text-heavy concepts.
- Goal setting and progress tracking throughout conversations
- Deeper personalization to match each student’s skill level and goals
Research is still ongoing to develop the most effective ways to apply AI in education. OpenAI is advancing further research on learning and artificial intelligence (AI) through various partners in its NextGenAI initiative, and is also collaborating with experts in the SCALE Initiative of Stanford University’s Accelerator for Learning (opens in new window) to study and share how AI tools impact learning outcomes in areas such as K-12 education.
In the long-term research on how students can learn most effectively with AI, OpenAI intends to publish a deeper analysis of findings on the relationship between model design and cognition, shaping future product experiences based on these insights, and collaborating with the broader education ecosystem to ensure AI benefits learners worldwide.