Anthropic has expanded the availability of its “learning mode” feature for the Claude AI, making it accessible to all Claude.ai users and introducing specialized versions for Claude Code developers and users. This move aims to transform the chatbot from a direct answer provider into a collaborative learning tool.
Initially launched in April for Claude for Education users, learning mode employs a Socratic method, guiding users to discover solutions themselves rather than presenting direct answers. Starting immediately, all Claude.ai users can access this feature through a new “Learning” option in the style dropdown menu. The experience mirrors that offered to educational users, allowing individuals to engage in a guided learning process, with the flexibility to deactivate the mode at any time.
For Claude Code, Anthropic has rolled out two distinct iterations of this feature. The first, “Explanatory” mode, enables Claude to generate summaries of its decision-making process during coding tasks. This provides users with insights into the AI’s logic, fostering a deeper understanding of the generated code. The second, more comprehensive option, also named “Learning,” is tailored for novice coders. In this mode, Claude will periodically pause and insert “#TODO” comments, prompting the user to write five to ten lines of their own code. Users can access these new output styles by updating to the latest version of Claude Code and typing “/output-styles,” which then allows them to select between the two modes or revert to Claude’s default behavior.
Drew Bent, Anthropic’s education lead, emphasized that the learning mode, particularly within Claude Code, represents the company’s effort to evolve its chatbot into a collaborative partner. “I think it’s great that there’s a race between all of the AI labs to offer the best learning mode,” Bent stated, adding his hope that this initiative will inspire similar advancements in coding agents.
The genesis of the original learning mode stemmed from discussions with university students who expressed concerns about “brain rot”—a term they used to describe the detrimental effects on long-term learning from simply copying and pasting chatbot outputs. Bent noted, “We found that they themselves realized that when they just copy and paste something directly from a chat bot, it’s not good for their long-term learning.” When adapting the feature for Claude Code, Anthropic sought to balance the needs of new programmers with those of seasoned coders like Bent, who possess decades of experience.
“Learning mode is designed to help all of those audiences not just complete tasks, but also help them grow and learn in the process and better understand their code base,” Bent explained. He envisions these new tools empowering any coder to become a “really good engineering manager,” meaning users might not write the majority of the code but will develop a keen understanding of how different components fit together and identify areas requiring further refinement.
Looking ahead, Anthropic is exploring further enhancements to expand upon the learning mode’s capabilities. As part of this ongoing development, the company is opening up Claude Code’s new Output Styles to developers, enabling them to construct their own customized learning modes. Additionally, users can now modify Claude’s communication style by creating their own custom prompts for the chatbot, fostering a more personalized AI interaction experience.
