Transcripts as Inputs

Hello world,

This week’s article is more of a lesson than my normal pontification.?It’s a quick how-to for using transcript data as an input to a GPT (like ChatGPT, Claude, Grok, etc). It’s super super useful, and really easy, and I’ve been lucky enough to find it, so here it is.?

How I learned it: My main contribution to the future of teaching and learning is making it easier to facilitate one-on-one conversations between learners. Because of this, I’ve gotten and given heaps of value from transcripts of live conversations, and my (our) work has led to millions of them.?It used to be hard. It has recently become easy.

What you can do with it (just a few ideas):?

(1) Record a conversation between you and a friend about some idea that you both understand pretty well. Transcribe it, copy and paste the transcription to a GPT and say something like, “Explain the idea these two discuss like a teacher would.” ?

Just do it in a new thread, before or after the transcript. The models are smart enough to infer what to do.

Now you’ve captured your idea!

(2) Host a mock interview with someone about a job they want to apply for and why they’re qualified. Transcribe the interview, copy and paste it to a GPT, and say: “write ____ a cover letter for the job they want.”

Don’t overthink it; just try it once - the ideas will flow from there. #educators especially, I can not wait to find out how you use this!!

Some logistics:

Transcription has gotten really easy. Apple devices come with a voice recorder app built in. That app transcribes all recordings. Android kids, you might need to open a Google doc or something. Apple can record and transcribe phone calls, the button is on the top left of the screen when you’re chatting. If you’re talking about something important or useful, consider transcribing!

For virtual collaboration, it’s just as easy. All platforms have a “transcribe” button these days. Google Meet, Zoom, etc.?Some people use fancy note-takers like Otter. I think that's overkill. Descript is super sick though!

Ok, lesson over, a little pontification before I hit the hay:

A little pontification:?

One of the hats I wear is of a precalculus teacher at my local high school. Precalculus teachers are responsible for ensuring that their students truly truly understand what mathematicians mean when they say the word function.?

Not everyone takes precalculus, and very few remember it, so here’s the main idea, with a little extra usefulness sprinkled in:

Functions (good ones at least) take abundant inputs and create useful outputs. In math class, it looks like taking some x and creating some y, like in y=3x+4, where the function takes some number x, multiplies it by 3, and adds 4 to it. AI is very “functional”. It takes some input, ie, a text-based prompt, and transforms it into something useful, hopefully. An image? Better text??

Good ideas are abundant. It’s far easier to chat about them than to write reports, blog posts, or things like this, etc. So, this article discusses the reasons for and process of using conversation transcripts as inputs for AI. They’re plentiful.

Written without editing or AI feedback (besides, like, Grammarly ).?