Claude won't just do your homework anymore

If you rely on AI to write code or explain complex topics, you already know the dirty secret: getting the right answer instantly often means learning absolutely nothing in the process.

On Thursday, Anthropic rolled out dedicated learning modes across both Claude.ai and its terminal assistant Claude Code. Instead of spoon-feeding solutions, Claude now flips into a Socratic tutor designed to walk you through the logic step by step.

A Socratic tutor in your chat box

On Claude.ai, the feature is available to all users right now. You just start a new chat, open the style dropdown, and select Learning.

Flip that switch, and Claude stops handing out finished homework. Instead, it asks guiding questions, probes your assumptions, and nudges you to unpack the core concepts yourself through what Anthropic calls guided discovery.

The timing isn't accidental. With the back-to-school season in full swing, tech giants are racing for students and educators. OpenAI just dropped its own Socratic Study Mode, and Google recently made Google AI Pro free for college students.

Claude Code leaves you with homework

Where Anthropic really got creative is inside Claude Code, its CLI and IDE coding assistant. Rather than treating code generation as a black box, Anthropic introduced two distinct learning styles.

The first is Explanatory mode. Here, Claude walks through its implementation choices, detailing why it picked a specific pattern and what architectural trade-offs it considered. Think of it as pair programming with a senior engineer who narrates their thought process.

The second is Learning mode, and it demands real effort. Claude writes the scaffold but intentionally stops at critical sections, leaving #TODO tags for you to fill in.

To try it out, update Claude Code to the latest build, run /output-styles in your terminal, and pick your mode. It marks Claude Code's tenth new feature shipped in the past month alone.

Why it matters

We think this is one of the smartest tactical moves Anthropic has made to counter the looming problem of AI-induced skill rot.

For months, the biggest knock against AI coding tools has been that junior engineers stop understanding the code they ship. By baking senior-dev explanations and intentional #TODO gaps directly into the CLI, Anthropic is turning an automation tool into an active sparring partner. For aspiring developers, this is genuinely worth switching on today.

The broader industry challenge is friction. Learning requires active effort, and when deadlines loom, most developers will naturally revert to default mode just to get tickets closed. The real test is whether Anthropic can make guided discovery convenient enough that people don't immediately turn it off under pressure.