> Why HN is so defensive against copilot but have a completely different opinion about MidJourney / StableDiffusion?
I've seen this question posted on HN so many times in defence of Copilot but I have yet to encounter people who fit what you're describing: people in general raise a lot of issues with MidJourney/StableDiffusion/&c. too.
HN is particularly defensive when it comes to Copilot because Copilot deals with sourcecode which is central to the daily vocation of a large portion of HN's readership. HN focuses on Copilot because they're subject matter experts in what Copilot trains on.
Also - as mentioned in siblings - while criticisms of training material sourcing by various ML models is common, there are quite a few details that makes Copilot's sourcing that little bit worse.
> it's inevitable and artists need to up their game
If this is your answer it seems you didn't even bother to read the title (never mind the article). There are two issues with Generative AI to discuss: input and output. Your answer addresses output - i.e. fear that what is produced may compete with humans. This article, and the vast majority of HN criticism of Copilot exclusively addresses input: objection to copyrighted works being used to train models without author consent. These are two entirely separate discussions.
I've seen this question posted on HN so many times in defence of Copilot but I have yet to encounter people who fit what you're describing: people in general raise a lot of issues with MidJourney/StableDiffusion/&c. too.
HN is particularly defensive when it comes to Copilot because Copilot deals with sourcecode which is central to the daily vocation of a large portion of HN's readership. HN focuses on Copilot because they're subject matter experts in what Copilot trains on.
Also - as mentioned in siblings - while criticisms of training material sourcing by various ML models is common, there are quite a few details that makes Copilot's sourcing that little bit worse.
> it's inevitable and artists need to up their game
If this is your answer it seems you didn't even bother to read the title (never mind the article). There are two issues with Generative AI to discuss: input and output. Your answer addresses output - i.e. fear that what is produced may compete with humans. This article, and the vast majority of HN criticism of Copilot exclusively addresses input: objection to copyrighted works being used to train models without author consent. These are two entirely separate discussions.