Got told AI did a better job than me.

[pmmg]

Which I think is Philip K. Dick's entire point. Love it. <3

A movie that really did this for me is Ex Machina. Its been out for a while so ppl have probably seen. But as it went on i started to question all of it. Who was a robot and who wasnt. I wish i could have that experience a second time but alas…. Ive seen it.

 

[skip.knox]

I recently submitted the bulk of a novel into Sudowrite (which uses a whole menu of LLMs) and basically asked whaddyathink? Most of critiquing was pedestrian enough, but there were a couple of surprising angles. For example, it saw a theme that I had not.

Which raises interesting questions in itself. A theme in a novel is a very abstract sort of thing. There are endless examples of Reader A seeing this them, while Reader B sees something else, and Reader C doesn't see a theme at all. At most, we can say that with some stories, the great majority of readers agree that this or that theme exists in the story.

The first interesting question is, what exactly did the AI notice? Recurrence, I suppose, is enough for now. There are probably layers and angles to ... oh I guess I'll call it Artificial Insight.

Now, the thing is, I take the AI's point. That really is a theme (I'm playing coy, not being specific) within my story. Plain as day, now that I've awakened. Now I start to look for opportunities to paint with that brush. I do believe it will add some nuance to the story.

Would an editor have noticed the same, since I'm claiming the theme objectively exists in the text. Unintended symmetry I suppose. Anyway, I think it would be much as with readers. One editor would say hey look at this, while another might not say a word about that theme, and yet another would perhaps suggest the theme is trite or unwanted. At that point, it's hard to prefer (on this one specific matter!) the human editor to the AI feedback.

Finally, as I add a sentence here or change a simile there, is that use inherently inferior or objectionable because I "used" AI? I am unable to see much difference between me being influenced by AI feedback and being influenced by an editor's suggestion or, indeed, my having read several books and by subconscious mysterious processes come to the same conclusion myself, happy in the illusion that I've been clever.

Gotta say, I'm loving the discussion as much as the tech.

 

[pmmg]

Finally, as I add a sentence here or change a simile there, is that use inherently inferior or objectionable because I "used" AI?

Yes.

 

[Ned Marcus]

Finally, as I add a sentence here or change a simile there, is that use inherently inferior or objectionable because I "used" AI? I am unable to see much difference between me being influenced by AI feedback and being influenced by an editor's suggestion or, indeed, my having read several books and by subconscious mysterious processes come to the same conclusion myself, happy in the illusion that I've been clever.

I don't understand the hostility towards using AI for things like editing. A good human editor or reader is better than AI, but AI is better than a mediocre critique group member or beta reader. AI is useful, why not use it?

 

[Mad Swede]

I've been staying out of this discussion, but now I'm going to be a bit blunt in my comments.

What you are saying skip.knox is that you do in fact need an editor, to help you spot themes etc that could be developed. This despite the fact that you have previously questioned what an editor could do for you... And Ned Marcus is saying much the same thing.

This is what my editor does, this is why I find working with her is so valuable. But my editor is also a person, with ideas, thoughts and feelings of her own. She adds her own touch of humanity to my work when she does her editing and my books benefit from that.

Which leads me onto my next point. In using some form of AI-tool for editing you are making several implicit assumptions about what that model has been trained on, asummptions which you cannot check and which you hope will not have a negative impact on your work. Those are not assumptions I as a published author would ever dare make, and they are not assumptions my publisher would accept.

 

[Zilver]

but AI is better than a mediocre critique group member or beta reader.

Is it? Again, I have no clue what people mean when they say 'better' in this context. Cause of what Mad Swede just typed out. Second all of that.

A beta reader tells me what impression the story made on them. AI could never do that, cause it experiences no impressions.

 

[Ned Marcus]

Which leads me onto my next point. In using some form of AI-tool for editing you are making several implicit assumptions about what that model has been trained on, asummptions which you cannot check and which you hope will not have a negative impact on your work. Those are not assumptions I as a published author would ever dare make, and they are not assumptions my publisher would accept.

I assume it's been trained on a lot of stuff, both good and bad. But none of that matters. What matters is whether it can help me or not. Does it give me an idea I can use or not? I see it as a brainstorming tool. As I said, I think a professional editor is better. Human proofreaders are unlikely to exist in the near future though. Nothing against proofreaders; I just don't see how they can compete against the online tools available now.

 

[Ned Marcus]

Is it? Again, I have no clue what people mean when they say 'better' in this context. Cause of what Mad Swede just typed out. Second all of that.

A beta reader tells me what impression the story made on them. AI could never do that, cause it experiences no impressions.

Better ideas.

On the second point, I agree. Beta readers (and critique group members) give a good idea of whether they like or dislike the story and what they like or dislike. If the beta reader is closely aligned with the type of storytelling you do, then I would not call them mediocre; I'd call them gold. But many are not.