Module · Detail

Prompting: Verify facts instead of just believing

This module develops the competence to formally distinguish between AI statements as facts or opinions and consistently equip facts with verification prefixes. Didactically central is the work with binding output formats that secure traceability and verifiability.

Time 12 minutes

The building block introduces a clear prompt structure that separates AI outputs into facts and opinions. The central modeling performance lies in establishing a binding output format (mandatory marking per sentence, fixed sentence count), thereby enforcing a verifiable text structure. By adding the rule "Verification possibility at [FACT]", the transition from assertion to justifiable statement is methodically secured. The connection between model and implementation is shown in that formal rules (label, bracket note, sentence count) directly control the response format and enable evaluation based on criteria. Thus, information quality is operationalized as part of reflective AI use. Compact

Language Art-Mode

Tasks 5

Introduction (Original excerpt)

If you need trust, let the AI visibly categorize each sentence: [FACT] or [OPINION]. For facts, it's important: don't just claim, but directly say how to verify it (without links).

Didactic classification

Subjective objective

The building block introduces a clear prompt structure that separates KI outputs into facts and opinions. The central modeling performance consists of establishing a binding output format (mandatory marking per sentence, fixed sentence number) and thus forcing an verifiable text structure. By adding the rule “prerequisite for [FAKT]” the transition from assertion to justifiable statement is methodically secured. The connection between model and implementation shows that formal rules (label, bracket hint, sentence number) immediately control the answer format and enable evaluation according to criteria. Thus, information quality is operationalized as part of reflective KI use.

Competence development

Create prompts with fixed formal requirements (sentence number, mandatory marking, additional rule for facts)
Classify statements as [FAKT] or [MEINUNG] based on linguistic features and verifiability
Formulate test cases for facts (e.g. study, survey, measurement) without referencing links
Control result formats according to criteria (Are all sentences marked? Does the classification of statement type match?)
Critically examine unclear or seemingly precise assertions and recognize them as warning signals (e.g. exact percentage values)
Correctly correct markings if the classification is not shared
Precisely phrase language so that verifiability and transparency become understandable in the output

Didactic added value in teaching

Structured progression through repeated application of the same prompt rules to different assertions
Clear task assignment with uniform formatting guidelines that facilitate comparison and control
Promoting systematic thinking by separating, classifying, and justifying statements
Transparent goal orientation: Verifiability is explicitly anchored as a criterion in the prompt
Support for differentiation based on the complexity of initial claims and the depth of test sets
Good connectivity to topics such as source criticism, argument analysis, and judgment formation

Flow of the lesson unit

Experiment 1: Chocolate before exams

This step establishes the basic structure of fact-checking by converting an everyday claim into a fixed sentence format with [FACT]/[OPINION] markings. The didactic focus is on adding a test possibility for facts to bind claims directly to verifiable criteria.

Didactic emphasis: Apply binding format rules for separation and verifiability
Typical challenge: Opinions are incorrectly marked as facts without a verifiable testing path

Experiment 2: Energy Drink and Math Note

This step sharpens the criteria work, because effect claims and grade references often sound like "fact" but are not verifiable without data basis. Didactically, the accuracy of classification is trained, and the quality of test sets (measurement, comparison group, survey) is worked out.

Didactic emphasis: Critically evaluate the verifiability of cause-effect relationships
Typical challenge: Causality is prematurely claimed, although only assumptions are present

Experiment 3: Learning with Headphones

This step focuses on the handling of subjective, vague terms (e.g. "epically enough"). The goal is to recognize linguistic markers for opinion statements and clarify the boundaries of factual verifiability in evaluative formulations.

Didactic Focus: Identify subjective criteria and mark them as opinion components
Typical Challenge: Ambiguities are overlooked, making statements appear as facts

Experiment 4: Plants in the Classroom

This step emphasizes the critical evaluation of seemingly exact numerical values that can be particularly plausible without context. Didactically, attention is drawn to verifiability, measurement definitions, and data foundations before an assertion is accepted as [FACT].

Didactic Focus: Use precision signals (numeric values) as a reason for examination
Typical Challenge: Accuracy is confused with truth; test setups remain too general

Experiment 5: Colorful Socks and Creativity

This step combines humorous initial claims with the demand for clear separation of fact and opinion. Didactically, it is practiced that even playful statements can contain verifiable partial aspects that must be clearly marked.

Didactic Focus: Secure sharpness even in pointed or humorous assertions
Typical Challenge: Exaggeration tone leads to blanket classification instead of examining individual statements

Task (Excerpt)

1. Write exactly 5 sentences about the given claim.
2. Begin every sentence with [FACT] or [OPINION].
3. Add a possible verification method or source of the information (without links) directly in parentheses for [FACT].

Example (excerpt)

This fragment is didactically relevant because it fixes the marking obligation and verifiability at facts as an output format that can be verified.

1
2
Jeder Satz beginnt mit [FAKT] oder [MEINUNG].
Wenn [FAKT]: Schreibe direkt dahinter in Klammern, wie man es prüfen kann (ohne Links).

Hints for classroom practice

Plan a short evaluation phase to check and possibly correct markings and verification criteria in a criterion-based manner.
Distinguish between the selection of statements (vague, causal, data-based) and additional requirements for verification criteria.
Secure results through a joint checklist (complete marking, consistent classification, concrete verification method).
Use extensions by highlighting individual sentences and justifying them or presenting alternative verification criteria.
Link to topics such as understanding statistics (number criticism), argumentation analysis, and reflective information evaluation.

This module makes facts and opinions clearly distinguishable through fixed format specifications and leads from statements to verifiable verification.

The clear task structure supports reliable result security through markings and verification criteria; differentiation is possible over varied statements and extension tasks.

Request a demo access and test the module in your own course context, adapted to your teaching goals.