I ran this on steve jobs 2005 video in stanford. It gave me wrong summary. The summary says Jobs gave 3 speeches at Stanford, in 2005, 2006, 2007. Please don't trust the chatGPT blindly. It has limitations as of now.

Video - https://www.youtube.com/watch?v=UF8uR6Z6KLc

Summary below:

>>. Steve Jobs delivered three speeches at Stanford University: one in 2005, one in 2006, and one in 2007. The 2005 speech is about trusting that the dots will connect in one's future, and how that gives confidence to follow one's heart. The 2006 speech is about how one's life destiny is something that can be trusted, and how that leads to happiness. The 2007 speech is about death, and how remembering that one will be dead soon allows one to make the biggest decisions in life with less fear.

I think adding Sponsorblock support so it doesn't look at sponsored fragments would be helpful. I tried one video and it confused the video topic with the sponsor.

I'm surprised by how fast/good these summaries are - well done!

I'm curious what prompt you used to generate them? If you don't mind me asking - I'm not sure how proprietary that is.

It only works with videos with CC but they could benefit from using OpenAi's Whisper API to get summaries of videos without captions

I love how GPT is weaving itself into so many different products.