As a part of my efforts to figure out how I could use AI tools, I decided to generate short summaries of some of my blog posts either with OpenAI API using gpt-3.5-turbo model or with ChatGPT using GPT-4 model.

I’m using a prompt similar to this one:

You are a network expert creating excellent short summaries highlighting the insights of boring articles. Your summaries will be no longer than three sentences or 100 words. Please create a summary for the following article written in Markdown

If the results are too boring, I might ask ChatGPT to create an exciting summary, and that sometimes turns its hype level to 11 ;)

Interestingly, most of the summaries are good or at least decent, and while GPT-3.5 generated a few that were totally off-the-mark, GPT-4 didn’t disappoint me (yet).