I came across an article by Suchir Balaji where he compares the chatGPT use of training data where he mentioned the reduced use of stack overflow after ChatGPT came into use.
although his work for to raise concern fair use of training data to create generative ai models, this article seem worth a read
https://www.nytimes.com/2024/10/23/technology/openai-copyright-law.html
P.S. Suchir Balaji is no more with us in exploration of AI research but has left us with his contribution that he contributed in collecting the data for gpt4. R.I.P.
Regards
DP