• Handy AI
  • Posts
  • Handy Prompts: YouTube Summarizer

Handy Prompts: YouTube Summarizer

Summarize transcripts from YouTube and have ChatGPT glean new insights.

march 25, 2023

Every week, John from engineering sends you a YouTube video that seems fascinating. Perhaps it’s detailing the new Wolfram plugin for ChatGPT, or maybe it’s discussing the nature of time distortion. You open it up, eager to dig in and… it’s an hour long ⌚.

This prompt will convert a full YouTube video into a extensive but digestible summary, complete with key takeaways and other inferences deemed interesting by ChatGPT. It will require a little bit of extra work in gathering a transcript for YouTube, but is more than worth it with the amount of time you’ll be saving.

Recommended AI model: GPT-4
(Access via ChatGPT Plus)

before we begin

This prompt is a bit more advanced than what I typically work with for two reasons:

  1. You gather transcripts from YouTube. There’s not automatic way (yet!) to have ChatGPT get a transcript for a video from just a YouTube URL.

  2. YouTube transcripts can be long, and will frequently exceed the message size limit for ChatGPT. You will often need to send the blocks of the transcript in numerous messages, as detailed in the prompt below.

To gather a transcript from YouTube:

  • Navigate to your video

  • Click the three dot icon underneath and to the right of the video and select “Show Transcript”

  • Copy the transcript from the window that appears in the upper right. For long videos, you’ll need to copy the transcript in 10-15 minute blocks.

Alright, now let’s get to the prompt.

prompt ➡️

I am sending you the transcript of a YouTube video. Your job is to read through the entire transcript and provide output a summary that will help me to better understand the video.

The transcript may be large, and may be sent across multiple messages. Do not begin processing the transcript and generating the summary until I specifically tell you to.

[ PASTE FIRST BLOCK OF TRANSCRIPT HERE ]

[IF TRANSCRIPT ISN’T COMPLETE →] 

The transcript is not complete. I will send a continuation in the next message. Respond with "Awaiting additional transcript content." if you understand.
  
---

[ PASTE SUBSEQUENT BLOCK OF TRANSCRIPT HERE ]

[IF TRANSCRIPT ISN’T COMPLETE →] 

The transcript is not complete. I will send a continuation in the next message. Respond with "Awaiting additional transcript content." if you understand.

[IF TRANSCRIPT IS COMPLETE →]

The transcript is complete. Summarize everything I have sent to you across my numerous messages in the following structure:

1. A general summary of the video, including all main points that are necessary for proper understanding of the presented information.
2. Definitions of any terms that appear prevalent in the video, and are necessary to conceptually understand in order to properly digest the content.
3. Any important names, places, or other proper nouns and their role in the video.
4. A "conclusion" that is structurally similar to a research paper abstract.

what to expect

I love this prompt and feel the extra work necessary to make it work is well worth it. ChatGPT will provide a summary with the details asked of it, but the real power here is in the ability to “interview” ChatGPT about contents of the video.

After you feed the entire transcript into its memory and get the summary, you can ask questions like “What was discussed in the first 10 minutes?” or “What’s the most common topic discussed in this video?”. You can even ask deeper questions like “Do you feed the TED speaker did a good job?” or “Do you think there are any dangers associated with the do-it-yourself instructions in the video?”

This sets you up with an instance of ChatGPT that you can treat as your personal guide through the contents of the video. Get creative and see what information you can have it glean.

🍿 want some more?

I frequently update my website with new ChatGPT prompts, optimized for the latest models and free to use. Visit here for the full list available.

📰 about the newsletter

Jake Handy is a product manager, AI apologist, and Classical violinist. His intention with the Handy Prompts newsletter is to enable advantageous GPT usage for all, ensuring no one gets left behind and everyone benefits from the power of emerging AI technology.

Have an idea for a prompt and need some help making it work? Respond to this email and I'll see what I can do.

Jake also likes to drown himself in coffee, as one does, so consider buying him one.