Getting your
Trinity Audioplayer ready...Have you ever used Google to find out how to do something? If so, the first several links are likely to be YouTube videos. I appreciate a good explainer video, but they take time to watch. Video is often best for hands-on tasks such as unclogging a sink, tying a tie, replacing windshield wipers or learning cooking techniques. Written instructions are usually faster and easier for things such as changing phone settings, fixing tech problems or following step-by-step directions.
Besides, YouTube creators have a tendency to waste your time. They often start with a “What’s up!” followed by a plea to subscribe and then a long-winded explanation of why you need to learn the very thing you’re there for.
Part of the problem is that much of today’s internet content is designed to maximize engagement, not efficiency. A simple answer that could be explained in 30 seconds is sometimes stretched into a 10-minute video because longer watch times can benefit creators.
There may also be step-by-step written instructions on the web, but they’re often harder to find. If you keep scrolling through Google results, you might eventually find them. Or now that AI is embedded into search, you might get AI-generated instructions at the top of the page. But AI sometimes gets things wrong, and it’s no substitute for expert advice, whether that comes from a written guide or a video created by someone who actually knows what they’re doing.
How to get a written version
But now you don’t necessarily need to watch. You can ask Google’s Gemini AI or ChatGPT to “watch” it for you and tell you what it says.
Here’s how on a web browser: Go to the video and copy the URL to your clipboard. A simple way is to highlight it in the address bar, right-click and select “Copy.” Then go to Gemini.google.com or ChatGPT.com and paste the video’s address as plain text by right-clicking and selecting “Paste as plain text.” If you paste without selecting plain text, it may insert the name of the video instead of the address. Before you hit Enter, type “summarize” or “summarize with time cues.” The latter shows you where to jump within the video to see the specific portion you’re interested in.
On a phone’s YouTube app, open the video, tap “Share,” tap “Copy link” and paste it into Gemini or ChatGPT.
An easier way
In some cases, you may see an Ask button in the row of buttons directly below the video. If so, click it and a Gemini sidebar will appear with options including “summarize this video,” suggested questions and a place to type your own questions. The Ask button doesn’t always appear, but when it does, it’s an easier way to query Gemini about that video.
Once you have the summary, you can ask questions that may be addressed in the video. If you want the answers to be based mostly on what’s in the video, it helps to start with something like, “Based on the video, what tools do you need?” Otherwise, you might get an AI-generated answer that may be useful but doesn’t necessarily reflect the expertise of whoever made the video.
Even though ChatGPT’s parent company, OpenAI, and Google are competitors, you can also use ChatGPT to summarize a YouTube video by pasting its web address and asking for a summary, time cues or anything else you might want to ask.
Going directly to AI
Of course, in the age of generative AI, you might not need to go to a video or any other web source. AI can answer questions directly and allow you to ask follow-up questions. As is always the case with AI, it can make mistakes. Never do anything that could cause harm without going to a trusted source. But for the most part, services such as ChatGPT, Gemini and Claude do a good job giving you instructions, including step-by-step directions. Before acting on AI information, I ask the AI for its sources and go to any that are listed that I know and trust.
**Transcribing audio files **
There are several ways to grab text from an audio file. Otter.ai and Descript specialize in that, but they charge a fee. Both the web and desktop versions of Microsoft Word have a good transcription feature. Click the down arrow next to “Dictate” (the microphone icon on the toolbar) and select “Transcribe.” Then upload your MP3 or WAV file. It will take a little while, but then you’ll get a full transcript that you can load into your document with or without speaker names and timestamps. The file will include ums, ahs and transcription errors, but you can run it through any generative AI product such as ChatGPT, Copilot or Gemini to get a cleaned-up transcript that’s usually true to the original audio.
Sometime that fastest way to watch a long video is to not watch it but let AI help you read or summarize what it has to say.
Larry Magid is a tech journalist and internet safety activist. Contact him at larry@larrymagid.com.