Just 'wrote' my first small snippets of #Python code for a #KNIME user question using the much hyped #ChatGPT (https://openai.com/blog/chatgpt/) and I must say I am somewhat impressed - the initial suggested code was already good but then 'we' started discussing trouble shooting options which were quite on point. Like having a coding-assistant (that one day might take over ....?).

I tried other things, some answers are very general and some links are broken (old training data).

https://forum.knime.com/t/batch-print-from-xlsx-to-jpg/56177/4?u=mlauber71

Introducing ChatGPT

We’ve trained a model called ChatGPT which interacts in a conversational way. The dialogue format makes it possible for ChatGPT to answer followup questions, admit its mistakes, challenge incorrect premises, and reject inappropriate requests.

@markks it's pretty good for everyday code, but once you get to less mainstream parts of the Python universe it starts to dream up code that does not exist. Which can be annoying if the imaginary code would be exactly what you need. 😅
@markks it's good to answer quite specific or very generic questions, and it'll certainly be able to take over a lot of the more pedestrian tasks over time (for now it still require supervision to avoid letting bugs slip in, but can't say humans are so different), but it's not able to build something complex out of the pieces at the moment (or ever?) so understanding what you are trying to build, which is the most important part of the job, is still quite necessary 😅 .

@markks So i kind of see it the same as a compiler, in a way, a compiler is taking high level code and transforming into code most devs can't or won't bother reading, but they need to know the high level instructions and how to make them work together. ChatGPT and others are able to move up the stack a bit, allowing you to think about bigger "instructions", closer to english, but still leaving you the design* of the app to do.

* for clarity, talking here about conception, not aesthetics.