Follow

As both a and a , I'm going to go out on a limb and say the ethical bar for is *far* lower than the bar for . Reading a bad story may make you feel like you've wasted an irretrievable chunk of your life, but it's very unlikely to land you in the morgue. So the idea that there might be ethical uses for in medicine but not in writing seems rather strange to me.

Jamming the two together: I'm alternately amused and annoyed by how much has doing things like , but relegates care to . Uh ...

Another addendum: like many writers, I'm a bit of a when it comes to the writing process. After decades of messing around with various word processors, these days I do my writing in using a text editor, then convert it to Word for sharing it with the rest of the world. I'm genuinely happier that way. checkers and the like can fuck right off.

But I'm uneasy with people who create rejecting the *idea* that might be useful for endeavors. I don't yet know what a good use case would be, and I doubt I'll ever want or one of its descendants to write a draft for me. Making it my work would feel like editing someone else's manuscript, even if it's good—which of course current AI writing isn't, but that could change. I just don't want people who think about the future professionally to assume that we'll never advance past the present.

@medigoth For me the issue is twofold. First, the environmental impact of our current AI (ChatGPT and the like) is atrocious! If we want to have a future we have to be far more sustainable.

Secondly, ChatGPT et al. is trained on stolen work. And honestly, in the few tests I've given it, I've found it to be completely, utterly lacking in creativity. I tried in both a business/marketing standpoint and in a more creative (help me brainstorm a name).

It may get better but right now it's rubbish.

@medigoth

I think the concern here is that using AI to write means using LLMs that have been trained on other people's copyrighted works, whereas using AI in medicine is about looking for patterns in complex medical data. Copyright isn't a concern in medicine, saving lives is.

Of course outsourcing medical advice to an LLM that has been trained on other people's copyrighted works sounds like an incredibly bad idea. There are probably more appropriate AI techniques for that, but people are blinded by the magic of AI you can talk to.

Sign in to participate in the conversation
Qoto Mastodon

QOTO: Question Others to Teach Ourselves
An inclusive, Academic Freedom, instance
All cultures welcome.
Hate speech and harassment strictly forbidden.