Editing assistance is where AI video earns its place today: auto-cutting long footage into clips, captions, reframing to vertical, and voiceover. Full text-to-video generation is improving fast but still reads as generated at longer durations, and platforms now require disclosure of realistic synthetic content.
Three different jobs, often confused
- Editing assistance — auto-cut a long video into clips, add captions, reframe to 9:16. Highest return, lowest risk, and what we teach first.Taught in our course
- Text-to-video — generate footage from a prompt. Genuinely impressive for a few seconds, still recognisably synthetic in longer cuts.
- Avatar and voiceover video — a presenter reading your script. Useful for explainers where nobody expects a real spokesperson.Taught in our course
What we teach students to do first
Shoot on a phone, then let the tool do the tedious part. One ten-minute recording becomes six clips with captions in the time it used to take to cut one.
Write the hook before you open the editor. Watch time is decided in the first three seconds, and no amount of transitions rescues a weak opening.
Caption everything. Most feed views happen with sound off, so the captions are the video for a large share of the audience.
Disclosure and monetisation rules
YouTube requires creators to disclose realistic synthetic or altered content in the upload flow, and applies a label to it. That matters commercially because YouTube's monetisation policies also target mass-produced and repetitive content, so a channel of pure generated output is a monetisation risk rather than a shortcut.
Meta labels AI-generated content and requires advertiser disclosure for social, electoral and political ads. The safe rule is the same as for images: if it looks like a real recording of a real person or place, say that it is not.