Gemini Guided Learning (Google)
Step-by-step Socratic mode in Gemini with retrieved diagrams, images, videos, quizzes and flashcards.
Gemini Guided Learning (Google)
- Maker: Google (Gemini app, LearnLM models)
- URL: https://blog.google/products-and-platforms/products/gemini/new-gemini-tools-students-august-2025/
- Status (10 October 2026): launched 6 August 2025, a week after ChatGPT’s study mode. Shipping in the Gemini app. Students in the US, Japan, Indonesia, South Korea and Brazil were offered a free year of Google AI Pro at launch. Closed source.
What it is
A Gemini mode, switched on from the prompt bar, that walks through problems step by step, asks questions and adapts explanations, aiming at the “how” and “why.” Launched with it: Gemini adds images, diagrams and YouTube videos to ordinary answers on its own when a topic is complex (the examples were photosynthesis and the parts of a cell), plus interactive quizzes, flashcards and study guides.
The problem it’s solving
Answer-giving chatbots let students skip the thinking; text-only answers are weak for visual, structural topics.
Its path / bet
Socratic behaviour plus retrieved visuals (existing images, diagrams and videos), and the separate Generative UI work (dynamic view) for generated interfaces. The visuals here are mostly found, not made.
How it works (concretely)
LearnLM-tuned Gemini models; image and video retrieval inserted into answers; quiz and flashcard generation from conversation or uploaded materials.
Strengths
- Combines the two learning ingredients with the best evidence: retrieval practice (quizzes) and text with relevant pictures (the multimedia principle).
- Found visuals (real diagrams, real videos) avoid generated-diagram errors.
Weaknesses / limits
- We found no published evaluation of learning outcomes for Guided Learning itself.
- Retrieved images are generic; they aren’t about your problem or your data.
Relation to fictty
Watch. A lab-scale bet on visual, interactive tutoring in a chat. Not a terminal or developer product. It’s evidence of where AI learning is heading (visual, quizzed), not a competitor.
Could fictty adopt it instead of building?
No.
What fictty should take from it
- Quizzes are cheap and they’re the part with evidence. Same lesson as Learn Your Way.
- Found before made. Where a real picture exists (a diagram in the repo, a chart a command already produces), show it rather than generate one. fictty’s image and SVG support should make “show the file” trivial.
Sources
- Google, New Gemini app tools to help students learn (6 August 2025)
- TechCrunch, Google takes on ChatGPT’s study mode with Guided Learning
- 9to5Google, Gemini Guided Learning