What happens when ChatGPT refuses to copy an author's style?

ChatGPT's got a new trick up its sleeve - it's now being pretty cautious about mimicking famous authors' styles. Instead of directly copying their voice, it's opting to capture the "broad qualities" of their work, all while keeping its own distinct tone. This shift could have some serious implications for users who rely on ChatGPT for writing tasks, and it's also got some big implications for the whole debate around copyright infringement and AI-generated content.

It's not entirely clear what sparked this change, but it's likely got something to do with the lawsuits currently pending against OpenAI, the company behind ChatGPT. Apparently, some of these suits are alleging large-scale copyright infringement, and one of them even specifically calls out ChatGPT's ability to generate text that's similar to copyrighted materials. By taking a more subtle approach to mimicking authors' styles, ChatGPT might be trying to cover its bases and avoid any more legal trouble. Still, it's a bit of a grey area - where exactly do you draw the line between capturing an author's "broad qualities" and straight-up copying their style?

This change is probably going to require some adjustments from users who rely on ChatGPT for writing tasks. Instead of asking it to write in the style of a specific author, they might need to get a bit more general with their prompts, or provide some extra guidelines to get the tone and atmosphere they're after. On the one hand, this could actually lead to some really creative and original writing - users will have to think more critically about the style and tone they want to achieve. On the other hand, users who've gotten used to relying on ChatGPT's mimicry skills might find this change a bit frustrating or limiting.

The implications of this change go way beyond just writing tasks, though - it's highlighting the whole tension between creativity, originality, and copyright infringement in the age of AI-generated content. As AI models get more and more sophisticated, it's getting harder to figure out what's "original" and what's "infringing". It's going to be up to users, creators, and lawmakers to navigate these complex issues and make sure that AI-generated content is both innovative and respectful of existing creative works.

Source: Ars Technica