Als nächstesNächster Leitfaden
Using AI in Group Projects
Anwendungen
Anwendungsleitfaden
Kids can use Scratch to build playful projects that model ideas behind AI, such as classification, recommendations and pattern recognition.
Standard Scratch code does not automatically learn from examples, so a project should clearly distinguish a programmed simulation from a model trained with a separate, approved tool.
Scratch is a block-based programming environment for making interactive stories, games and animations. It is a good place to explore how people design systems and explain an idea through a working project. A Scratch project can demonstrate AI concepts, but ordinary Scratch code follows the blocks a student writes; it does not learn from examples simply because the project contains a character or a clever decision. Start with a question a child can explore: How might a game sort recyclable objects? What clues could help a character recognize a sound? Which examples might confuse a rule? A student can program a transparent decision tree or a set of if-then rules, then test edge cases. Call it a rule-based simulation, not a trained model. To explore machine learning directly, pair Scratch with a separate, school-approved learning activity or extension that actually trains a model, and identify that component clearly. Plan for iteration. Students should choose labels, create varied examples, test cases the system has not seen and note mistakes. Ask what changed when they added or corrected an example. Discuss how narrow examples can produce a narrow result and how the user should respond when the system is unsure. Scratch Team starter projects can help beginners remix animations, games, stories and interactive art; remixing should include attribution and a clear explanation of the student’s changes. Keep young creators safe. Use school-approved accounts and tools, avoid entering names or identifiable images of classmates, and check sharing settings before publishing. If a camera or microphone is involved, explain what is processed, obtain required permission and prefer an activity that keeps data local when possible. Assess the student’s explanation, testing and revisions as well as the finished project. The learning goal is to make a system understandable and question its limits, not to label every interactive program as AI.
Das Design auf Anwendungsebene bestimmt, ob KI tatsächliche Ergebnisse verbessert.
Eine gute Workflow-Integration führt zu Produktivitätssteigerungen, denen Benutzer vertrauen können.
Gut abgegrenzte Anwendungsfälle reduzieren die Änderungsmüdigkeit und das Implementierungsrisiko.
Block-based tools may connect more easily to sensors and machine-learning services, letting children compare a hand-written rule with a learned pattern. That makes careful labeling and privacy instruction more important. Teachers should check that an extension is maintained, approved and understandable before students use it, then help learners explain where data goes and how to correct a mistake. Scratch can remain a creative coding space whether or not a project uses a trained model. The lasting skill is being able to describe what the program does and what it cannot know.
Make a Scratch sprite sort objects by a student-written rule, then compare the rule with how a data-trained classifier might behave.
Build an interactive story in which a character asks questions and follows a decision tree the student drew.
Use labeled example cards to discuss how a classifier might confuse two categories before implementing a simple rule-based version in Scratch.
Pair a Scratch animation with an approved, teacher-supervised machine-learning activity and explain which part is the trained model.
Die Automatisierung eines fehlerhaften Prozesses kann bestehende Probleme verstärken.
Teams können zu stark automatisieren und das notwendige menschliche Urteilsvermögen verlieren.
Die Qualität kann schwanken, wenn die Ergebnisse nicht kontinuierlich bewertet werden.
Ordnen Sie den aktuellen Arbeitsablauf zu und identifizieren Sie den Schritt mit der höchsten Reibung.
Definieren Sie menschliche Kontrollpunkte vor der vollständigen Automatisierung.
Schulen Sie Benutzer in Bezug auf Eingabeaufforderungen, Eskalationspfade und Qualitätsstandards.
Verfolgen Sie Ergebnisse auf Aufgabenebene, um den nachhaltigen Wert zu bestätigen.
Free newsletter
Three verified AI stories every weekday morning, written in plain English. Free forever, no ads.
One email each weekday. Unsubscribe in one click. We never sell or share your address.
Test yourself
Instant feedback on every answer, and a shareable certificate with a verifiable ID once you pass a course.
Support free AI education. AI Understanding is a 501(c)(3) nonprofit — no ads, no paywall, ever. Make a donation
Kids can use Scratch to build playful projects that model ideas behind AI, such as classification, recommendations and pattern recognition. Standard Scratch code does not automatically learn from examples, so a project should clearly distinguish a programmed simulation from a model trained with a separate, approved tool.
Ordinary Scratch conditionals execute programmed rules rather than learning patterns from data.
The student needs to identify a real training component rather than imply that Scratch itself learned.
Held-out examples can reveal whether a learned mapping works beyond the training set.
A system may fail on cases that differ from its limited examples or conditions.
Remixing should preserve appropriate attribution and make the student’s contributions clear.
Lerne weiter
Weitere Leitfäden zu diesem Thema ausgewählt
Als nächstesNächster Leitfaden
Using AI in Group Projects
Anwendungen