How To Make YouTube Shorts With AI (Step-by-Step 2025)
Core: YouTube Shorts can be created quickly using AI tools like Descript through three methods—generating videos from scratch with text prompts, editing your own footage with AI assistance, or repurposing existing long-form content into viral clips.
One-line summary
This tutorial teaches how to leverage Descript's AI-powered features to create YouTube Shorts through three distinct workflows: fully AI-generated videos, AI-assisted editing of personal footage, and automated repurposing of long-form content.
Prerequisites
- A Descript account (preferably a Creator plan or trial to access core AI features)
- Basic understanding of YouTube Shorts requirements (e.g., portrait aspect ratio, maximum duration)
- (Optional) Existing video footage or long-form content for methods 2 and 3
Step-by-step
Step 1: Start a new AI Video Maker project in Descript
Open Descript (either web browser or app) and select the "AI video maker" option from the new project presets. When presented with options for getting started, choose "Generate from a prompt" to create a video purely from text input without uploading any media.
Why this matters: This initiates the fully AI-driven creation workflow, allowing you to build complete videos without recording anything yourself.
The interface presents clear options including "Upload a file," "Generate from a prompt," and "Paste in a script," making it straightforward to choose your preferred starting point.
Step 2: Provide a text prompt and customize AI speaker/avatar
Enter your video idea as a text prompt (e.g., "YouTube short on top video recording tips"). The AI will use this prompt to generate an entire script tailored to your topic.
Next, select an AI speaker voice from the available options. The demo uses "Owen" as an example. You can preview different voices before making your selection. Optionally, you can assign an AI avatar or use your own trained voice/photo if you've previously cloned your voice or uploaded custom avatars.
Why this matters: The voice and avatar selection determines the personality and presentation style of your video, making it crucial to choose options that align with your brand or message.
Step 3: Set aspect ratio and choose media type
Select "Portrait" for YouTube Shorts to ensure the correct vertical format. Then decide whether Descript should use "AI generated media" (animated images) or "Stock media" (stock video clips) for the B-roll and on-screen visuals.
Why this matters: Stock media is generally recommended for video clips to create a more dynamic look, while AI-generated media may result in animated still images rather than video footage.
AI generated media may result in animated images rather than video footage.
The interface shows various illustration styles and provides a preview of how your content will appear in the chosen format.
Step 4: Generate, review, and edit the video
Allow Descript to generate the video, including script, voice-over, and visuals. The process takes a few moments as it divides the script into scenes and generates appropriate content.
Once ready, review the generated video. The interface displays your script on the left with synchronized visual output on the right. You can:
- Edit the script by typing directly into the text
- Delete unnecessary sections
- Replace B-roll clips with other stock media by selecting clips and choosing "Replace media"
- Customize captions (font, size, color)
When satisfied, export the final video locally or upload directly to YouTube through Descript's built-in publishing options.
Always review the AI-generated content for accuracy and brand consistency, as some elements may need manual adjustment.
Why this matters: While AI handles the heavy lifting, human oversight ensures the final product aligns with your specific goals and maintains quality standards.
Step 5: Start an AI Video Maker project and upload your footage
For creating Shorts from your own raw footage, choose "AI video maker" and then select the option to "upload an audio or video file."
Upload your raw video footage (e.g., a widescreen recording) that you want to transform into a Short. This could be unedited talking-head footage, product demonstrations, or any other content you've recorded.
Why this matters: This method allows you to maintain your personal presence and original content while still benefiting from AI-powered editing and formatting.
Step 6: Configure short format and B-roll options
Set the output format to "Portrait" for Shorts. Choose "Stock media" for B-roll to allow Descript to find relevant clips to accompany your footage, enhancing the visual appeal.
The AI analyzes your uploaded content to understand the context and select appropriate supplementary visuals.
Step 7: Review and refine the AI-edited video with Underlord
Descript will automatically convert the footage to portrait format, clean audio, add titles, captions, and B-roll, and attempt to remove mistakes. Review the auto-edited video to see what the AI has created.
For advanced mistake removal, enable the "Underlord beta" feature in settings under the Labs section. Once enabled, use the Underlord command with a prompt like "Only use the last take" to ensure only the best takes are used and all bad takes, mistakes, and restarts are intelligently removed.
The Underlord feature processes your request and removes redundant takes without cutting sentences in half, prioritizing only your final, successful attempts. This dramatically reduces editing time compared to manual take selection.
Fine-tune layouts, titles, and B-roll as needed. The AI automatically crops and centers the active speaker for portrait format, but you can adjust positioning and scaling if necessary.
The Underlord beta feature is still in development and might require manual oversight to ensure desired results. Remember to check settings for beta features.
Why this matters: Underlord represents a significant advancement in AI editing, automating the tedious process of identifying and removing bad takes while preserving the natural flow of your content.
Step 8: Import long-form content into Descript
Import an existing long-form video, such as a live stream or podcast, into a new Descript project. Simply upload the file as you would with any video editing project.
Descript will automatically transcribe the content and identify different speakers, preparing it for repurposing. The transcription appears alongside your video timeline, making it easy to navigate through the content.
Why this matters: This prepares your content for the powerful "Create Clips" feature, which can extract multiple viral-worthy moments from a single long-form piece.
Step 9: Use "Create Clips" AI action
Navigate to "AI actions" and select "Create Clips (Viral bait while you wait)".
Configure the settings:
- Specify the number of clips you want (e.g., 10)
- Set a guideline duration (e.g., 60 seconds)
- Optionally provide context or criteria for finding clips to focus on specific topics
The AI analyzes your content to identify the most engaging, "viral-worthy moments" and automatically generates multiple short clips.
The duration guideline is flexible; clips may be slightly shorter or longer than specified.
Why this matters: This dramatically reduces the effort required to extract highlights from long-form content, with the AI consistently identifying engaging moments better than many manual approaches.
From testing many AI tools for content repurposing, Descript's "Create Clips" is one of the best at consistently finding good clips that work in other contexts.
Step 10: Review and customize the generated clips
Descript will quickly generate multiple short clips, providing titles, descriptions, and a virality score for each. These clips are saved as sub-timelines within your project.
Review each clip by opening it from the project compositions. For each clip:
- Adjust the speaker's position and scaling for portrait format using the scene editor
- Customize titles and captions to match your branding
- Modify layouts to ensure visual consistency
Save custom layouts as templates for future use, allowing you to apply consistent branding across all your Shorts with one click.
Auto-generated layouts may not perfectly align with branding; manual adjustments and template creation are crucial for consistency.
Why this matters: While AI handles the heavy lifting of identifying and extracting clips, customization ensures each Short maintains your unique brand identity and professional presentation standards.
Common mistakes & fixes
Not fully leveraging Descript's advanced AI features for editing or content repurposing, leading to more manual work. Explore and utilize features like "Underlord beta" for refining takes and "Create Clips" for quick content repurposing to maximize AI efficiency.
Using AI-generated media for B-roll when actual stock video clips are desired. Ensure "Stock media" is selected when prompted if you want dynamic video footage for B-roll rather than static, animated images.
Not customizing the auto-generated content to fit specific brand guidelines or desired messaging. Always review and make necessary adjustments to scripts, B-roll, captions, and layouts, as AI outputs often need human refinement for perfection and brand consistency.
Expected result
Professionally edited and engaging YouTube Shorts, created quickly with significant AI assistance. The shorts will feature appropriate aspect ratios, B-roll, captions, and polished audio/video, tailored to the chosen method (from scratch, from own footage, or repurposed long-form content).
The final videos should be ready for upload to YouTube with minimal additional editing required, demonstrating professional quality while saving significant production time.
Going further
After mastering these three methods of creating YouTube Shorts with AI, consider:
- Exploring voice cloning in Descript to use your own voice for fully AI-generated videos
- Learning advanced caption styling to create distinctive branded text overlays
- Developing a content repurposing strategy to systematically extract Shorts from all your long-form content
- Investigating YouTube Shorts optimization techniques to maximize engagement and views
- Creating Descript templates for consistent branding across all your AI-generated Shorts
- Studying short-form video storytelling to craft more compelling narratives within the 60-second format
Related concepts
- Descript
- AI video editing
- YouTube Shorts strategy
- Content repurposing
- AI voice cloning
- Stock media libraries
- Video transcription
- Automated video production
- Portrait video format
- Social media video optimization
Key timestamps
- 00:00 — Introduction to making YouTube Shorts with AI using three methods
- 01:06 — Overview of Descript and its sponsorship
- 02:18 — Method 1: Creating a video from scratch with AI
- 02:50 — Demo: Generating video from a text prompt in Descript
- 03:52 — Customizing AI speaker and avatar
- 04:36 — Choosing aspect ratio and media type (stock vs. AI-generated)
- 05:30 — Reviewing and editing the AI-generated video
- 07:56 — Method 2: Creating a video from your own footage with AI
- 08:25 — Demo: Uploading footage and letting AI edit
- 09:37 — Reviewing AI-edited footage and using Underlord beta
- 12:50 — Method 3: Repurposing existing content with AI
- 13:30 — Demo: Using "Create Clips" for viral moments
- 14:50 — Reviewing AI-generated clips and customizing layouts
- 16:35 — Conclusion and call to action for Descript