I wanted to try out ChatGPT o1 model to generate a full streamlit app, asking Fermi style of estimation questions, without looking at the code

With some simple prompts, it did a fairly good job, in about one hour

I only needed to copy paste error messages a couple of times, for it to suggest corrections

My take-away is that LLMs will empower many non-technical users to achieve their (simple / self-contained) goals with no need for external help, which makes me hopeful about their benefits to society