• You are responsible for the content, not the AI.
• Language models often generate plausible but factually incorrect statements („hallucinations“).
• Legal uncertainties (copyright, data protection, risk of deception) often arise from the uncritical sharing of AI content.
• Good review is not a check „out of mistrust,“ but a professional quality standard.

Content Review

Ask yourself: Is this correct—really?

• Are all facts and figures correct?
• Is a source cited—or does it just sound like it?
• Are there exaggerations, contradictions, or outdated statements?
• If you see legal or technical statements, ask the AI a follow-up question: „Please provide the source for this statement.“ This way, you can verify if it is well-founded—or just „sounds good.“

Formal Review

The text may be correct—but does it fit?

• Is the structure clear?
• Is the format correct (e.g., length, outline, relevance to the target audience)?
• Has the required text type been implemented (e.g., invitation, strategy paper, concept)?

Stylistic Review

Does the text appeal to you—or does it sound like a generic AI draft?

• Are there phrases like: „In the digital age, innovation is paramount“?
• Do words or sentence patterns repeat?
• Is the tone factual, motivating, professional—or somewhere in between?

If you used your style prompt from Module 2.3, the tonality should already be quite good. Nevertheless, read it over to see if you recognize yourself in the text—or rather an anonymous AI author.

Impact Review

Does the text fulfill its purpose?

• Does it inform? Does it persuade? Does it activate?
• Does the tonality fit the target audience and the channel?

Before you start optimizing, you need to know: Is this even correct?

• Ask follow-up questions to the AI, e.g.: „On what is this statement based?“ „Are there studies on this?“
• Watch out for apparent certainty („It is well known that…“)—without proof, it is not reliable.
• If you are unsure: Search for a source yourself or ask the AI to provide sources (including URL or author).
• Especially with legal statements, diagnoses, and scientific claims: Always double-check.

AI texts should not only be fact-checked—but also reviewed in terms of their societal impact. Check:

• Is it clear that the content is (partially) AI-generated? → You can find more on transparency obligations in Module 1.5.
• Is a distorted worldview being conveyed? (e.g., stereotypes, unbalanced language)
• Is diversity considered? Are different perspectives possible?
• Are people or groups being discriminated against—intentionally or unintentionally?

AI models are often trained on „majority opinions“—this does not automatically make them fair or nuanced.