Chapter C12 · Interactive Course
Images, voices and trust
Judge generated media, and refuse to be hurried.
Core Learning Concept
Inspect a surprising piece of media before sharing it or acting on its message. Separate what it appears to show from evidence about its origin. Look for the original publication and independent confirmation. For a consequential request, contact the supposed sender through a channel you already trust.
What You’ll Learn in This Chapter
-
AI can generate visual content
Recognize generated imagery as created output rather than proof of a photographed event.
Image generation creates visual content from an input such as a description or reference image. The result can depict an event that never happened. Treat it as created media unless its origin establishes otherwise. Recognizing what generation can do helps separate visual realism from evidence of a real event.
-
Describe what to show
Give a clear subject and intended use for a simple image request.
An image brief describes the subject, setting, arrangement, and details that matter. “A poster showing three books beside a reading lamp, with space for a title” is easier to judge than “make it nice.” Check the output against those requirements, including any text, quantities, or factual details.
-
Inspect the details in an image
Check generated visual details against the task.
Always inspect generated images for critical task details: spelled text, correct dates, and object counts. An image can look beautiful overall while containing gibberish spelling or wrong dates. Zoom into key areas before using generated graphics.
-
Voices and video can be generated too
Recognize that familiar-sounding media can be synthetic or altered.
AI can generate or transform voices and moving images as well as still pictures. A familiar-sounding voice therefore does not by itself establish who is speaking. Separate the media’s appearance from its origin and verify consequential requests through an independently trusted contact method.
-
Spotting glitches is not a proof test
Avoid definitive authenticity judgments based only on appearance or detector labels.
Visual glitches like strange fingers or warped text are useful clues, but their absence does not prove an image is real. Generative tools improve constantly. Treat visual appearance as only one hint, and look for independent evidence and original sources.
-
Trace where media came from
Use origin and edit-history information without confusing it with factual truth.
Provenance is the verified record of where media originated and how it was edited. Finding the earliest publication and creator is far more reliable than guessing from visual style. A real photo can easily be miscaptioned to deceive.
-
Verify a surprising request
Confirm an urgent request through a known separate channel.
A surprising urgent request needs identity verification before action. If a voice message asks for a consequential change, contact the person using a known number or established channel. Do not rely only on contact details supplied within the suspicious message, because those could lead back to the same sender.
-
Would you share or act on this?
Review a poster and an urgent voice message: correct a visual detail, distinguish provenance from truth, and choose independent confirmation.
-
Check before sharing
Audit an event pack containing a poster, a written recommendation and an urgent voice-message transcript before deciding what to share or act on.