CHATgpt to help with scripts

[RyanCameron]

Anyone else try to use ChatGPT to help them with API or Dynamo scripts? Granted the verbiage was slightly off for the family.load node, (family types) it still worked. Is this a game-changer for anyone else?

[jw.vanasselt]

Yes, it is.

It does not always give the right solution but it points you in the right direction.

[ricaun]

I give some tests in the ChatGPT in January, and Revit API is good when you ask for simple stuff. If you ask for something more complex or something that does not exist in Revit API, oh boy, ChatGPT creates methods and proprieties that do not exist in the API and some of them should be really useful if exists.

I found really useful to use to refactor methods and documentation, or even create unit tests for some code.

[RPTHOMAS108]

You could have got that from an internet search and that process would have also forced you to consider the individual parts of the problem and what they could be used for in other forms. You just seem to be doing it parrot fashion.

It is not good for learning, humanity will not in the end benefit from it. It is all quite depressing how people are subcontracting their individual intelligence out to AI these days. I can understand how AI could be used for big picture problem solving but not 'how do I do my homework'.

The development of AI in that form will have an inverse relationship with general human intelligence level. Perhaps 'intelligence level' isn't the right phrase when I actually mean 'problem solving ability'. It is the dependency on it that will create the problems i.e. what happens when it is manipulated by the small amount of people that understand how to put bias etc. into it? The people as a whole will believe it even when it isn't being truthful.

In this context here it doesn't matter because you'll just get it wrong and realise it has mislead you. If you do a history assignment however then suddenly the history is what the system knows about and through popularity of use it becomes the perceived truth. Who or what is the arbiter of truth anyway, those texts that can be easily transcribed into a data set?