I couldn't replicate his results about "reasonable prompts" with ChatGPT 3.5. I did a small write up and a critique here: https://www.linkedin.com/posts/terokeskivalkama_so-some-tips...