Let me assure you that the idea of AI disgusts me. Not just because it debates the quality of work and is a “cheat” to get past the word target, but because it stripey the writer themselves of their accomplishments.
However, it does strike me that people might want to write about subjects that they don’t know a lot about. Scientific knowledge, crime, historical stuff. Could that be an ethical place to use AI - not to write it, but to ask “what would confectionary in the 1300s look like?” (For instance). Just niche questions.
Or is there a much better, easier way to look up niche real-world details like that?
If we were to close our eyes and pretend that the conversation around AI hadn’t happened the way it did, and the training had only involved things people generally agreed were fair game, one could imagine a world where we largely used them as “that person who can tell you what to search for” or clarifies your understanding.
It’s not intelligent in the way anyone means, but it’s theoretically capable of being a better interface for a search engine.
Edit: it’s whole thing is language manipulation. I forgot that I’ve written technical documents and had it summarize or explain them back to me. Anything it gets wrong is something I can bet a lazy or rushed human would also get caught by, so I clarify that point. Not proofreading, since I’m not asking for feedback. Just “did the words convey the information well enough for a statistical model to properly contextualize”.
it’s whole thing is language manipulation
That’s a good point - it’s worse than searching something in a browser, so just search in a browser
Yup. Where it gets fuzzy is with this type of AI being integrated into a search engine.
I honestly feel like if Google had waited, made their solution a good interface for searching as opposed to something which tells you answers and then finds references as best it can that it would have been a lot better. “We trained our model on the same data that our search engine looks at” is a lot more palatable than where things are now.
The tedious part of research for me is figuring out the keywords I need to get to meaningful information.
Something that can say “what you described is called ‘ball and socket modular hose’, here are some good starting points:” would actually be useful. (I just went through that rigamarole trying to find those rigid little flexible water or air nozzles that can snap together, so it’s fresh in my mind)
No.
What about an art contest for AI generated graphics
If you ignore the environmental impacts and how you’re making rich assholes even richer, it could maybe be used for editing advice (with a huge grain of salt for hallucinations, but it could sometimes point out something you hadn’t noticed).
But you should not ignore the environmental and economic impacts. Instead, you should become a better writer and find human pre-readers.
(Oh yeah, and it’s not strictly an ethical concern, but if you use it for editing, you need to be concerned about sending it your in-progress work, as they will likely be stealing everything you send to use as training material and launder into their own ‘original’ work.)
Oh yeah, and it’s not strictly an ethical concern, but if you use it for editing, you need to be concerned about sending it your in-progress work, as they will likely be stealing everything you send to use as training material and launder into their own ‘original’ work.)
I haven’t done this, but I do wonder how far it goes. Like if we type something in Google docs without ublock origin, it technically gets used for LLM training. But i don’t think I’ve yet heard of someone’s work or personal details being lifted and used by an AI elsewhere - have you heard any examples of that?
And then again maybe the ad+tracking blockers i use don’t work
Post-apocalypse with a system that got saved over in a lead lined locker.
Alan Moore’s From Hell would probably need another two books just for him to talk about his research and the references he makes and it’s just a comic book… bad readers can hardly be good writers.
I guess it would be fair to ask LLMs to start your journey, like “main works on this genre/theme”, and from these works you pull the thread for more things
I guess it would be fair to ask LLMs to start your journey, like “main works on this genre/theme”, and from these works you pull the thread for more things
On reflection, I guess the safest way to use it is to say “find me a source on x, and y” and look them up, instead of asking it direct questions
Alan Moore’s From Hell would probably need another two books just for him to talk about his research and the references he makes and it’s just a comic book… bad readers can hardly be good writers
I’ll bear that in mind then.
Well, the ethical side of things is more about how LLMs were made, rather than what they do. However, you would also need to disclose what use you made of even the most ethically built LLM.
That being said, the specific use case you’re positing is bullshit. You can’t ai away research. Worse, it is likely to backfire because GIGO.
What you’re describing is just a fancy search engine, and you can’t trust any of the ai driven ones to give good information reliably. Hell, it’s hard enough to use a search engine to do research now, because mass generated shit articles clog everything up. Using ai to parse ai results is just absurd.
If you aren’t capable of running a search on your own, chances are high that you won’t be able to turn anything into a useful story anyway. There’s a degree of competency to writing. It’s called “craft”, and no amount of ai can bypass it if you want to write. So learning how to parse bad search results into useful information is part of the learning curve. Either that, or you learn the Dewey decimal system.
Edit: to answer the title question: editing. Self published writers simply don’t have access to editors and copy editors. A good, locally hosted, LLM would be able to fill that gap
Or is there a much better, easier way to look up niche real-world details like that?
Yeah, a search engine…
The first one was invented over 35 years ago.
Most of them, just read the first Wikipedia entry
https://en.wikipedia.org/wiki/Confectionery_in_the_English_Renaissance
After reading about it, you’d recognize some names like “fritter” and already have an idea of modern versions to visualize.
Then you can search for modern recreations of those recipies if you want to actually see what they look like. Because if you meant image generation. You’ll get slop you can’t and shouldn’t trust. You’ll never know if it’s accury
Think of it like a flamethrower, extremely powerful and extreme dangerous, not for any kind of subtle work, but if you need to blast a field of weeds it’s the best way to do it.
AI is destroying the planet and 100% based on the work and knowledge of individuals. Most of them haven’t agreed or even been asked if is okay to make profit of their work.
There is NO ethical use of A.I.
Every decade of my life, new technologies for doing creative work have emerged that purport to reduce the workload of creatives yet always produce an inferior product. 3D CGI in film supplanted hand-drawn animation and now most kids movies look like shit. Digital special effects and motion capture supplanted practical effects, and now films don’t hold up visually at all. Now we have the idiot box, and not only does it turn the written word from gold into lead, the effects of its use have frequently lead to decreased intelligence, more psychosis, placated personality disorders, and just straight-up fascist misinformation.
But, I am a permissive critic, and I’m always interested in what motivates a creative choice over whether a work is merely entertaining or pleasant. If Satan himself took a burning shit in your hand and you fingerpainted with it, I would be spellbound if you managed to do something interesting and meaningful with it; most of us would just scream in pain and jump from a window without delay. That’s the beauty of art, don’t you think?
Idiot box used to be the TV.
The issue with CGI is a bit complex. We see it all the time in ways that are completely unnoticeable. And the reason CGI is so bad is partially an artifact of producers refusing to stick to a vision; they are constantly asking for changes and digital effects houses can’t tell them to go fuck themselves if they want to ever work again. It’s an absolutely brutal business. The CGI house that won an Oscar for the Life of Pi went bankrupt the next year. So what you’re seeing with bad CGI is probably the crunch work of the week before opening day. That was the story of The Flash.
That said, CGI is often just the cheapest tool for the job, but often not the best.
Two videos that cover this: No CGI is just invisible CGI: https://www.youtube.com/watch?v=7ttG90raCNo
and
a comparison of Carpenter’s The Thing (practical) and the remake. Thesis: we know the practical effects are fake, but we know the CGI is not real: https://www.youtube.com/watch?v=7ttG90raCNo
I always forget this about CGI studios, the economic side of it. And yeah, I know CGI can be done well but it so rarely is partly because of that meddling.
I should have come up with an original pejorative for LLMs, calling it idiot box was a sleepy mistake.
A lot of what I said last night was motivated by a fresh rewatch of Hellraiser, admittedly. Even the hokey lightning and magic VFX had a sort of handmade charm to them, everything held together really nicely even after nearly 40 years.
The The Thing video I linked talks about how practical effects are more viscerally effective, even if they look “bad” our brain registers them as “real” at least; while our brains don’t give any weight to things that it recognizes as CGI.
I’ll make a point to watch that one in my downtime, then. I wasn’t even aware of a Thing remake until you mentioned it, but even knowing it exists I’m going into it already slightly irritated. Practical effects are hefty!
All this to say, even though my original point was kind of badly put (tone doesn’t carry well in text, I’m basically on the same page as everyone else), what you do with a tool like CGI can’t just be cheaper or easier for a studio exec to meddle with, it has to be equally or more interesting than doing the work by hand or it just sucks.
Like, I know studios aren’t purely motivated by artistic integrity and never were, but the more we put new tools for cheapening the output into their hands, the worse our art gets. I think the same holds true for when small independent artists get those same tools. Some people will use them in unique ways that add to the conversation and bring culture forward, but most will just use it to make cheaper stuff. My fingerpaint/jumping out the window analogy didn’t really make that clear, that’s on me. I just think that doing things by hand is always going to (generally) work better than any tool that makes doing the work ‘easier’…
At least until we have removed the cost of doing art from the equation of what gets made. Once we’re in a post-scarcity mode I’ll be all for more powerful tools that make production easier.
If you’re really wanting to evaluate the impact/benefits of CGI en toto, watch that first video. I think it points to there being some things CGI is really good for–cost effective and invisible–and things it is bad for.
I think it’s about figuring out where that uncanny valley is. Re-watching LOTR, pretty much anything to do with people and people’s movement is really hard to get right. Like the shots of them crossing the bridge at Moria were composited live-action shots syncing cameras between the miniature set and the live actors. But lots of Legolas’ CGI action sequences are comically bad.
But let’s keep in mind that a lot of great classics, like The Mummy (Fraser) and T2 and The Abyss all were made with primitive CGI and still look fantastic, so it really is about approach.
There is also a new thing that directors have to worry about–whether their carefully planned, all-day-setup-single-take practical effect will be dismissed as CGI. There are practical stunts that are not worth doing anymore because people will assume it’s CGI.
But, all in all, between the bad CGI and the oppressive color grading, this most recent era of movies is not going to age well at all.
Agreed. I think part of the creative choice you have to make with practical effects is, if you’re going to do practical effects, do something so crazy people would assume was CGI and then brag about how it wasn’t.
Fury Road used CGI to great effect because of how intelligently it was used, like you say - invisible. Putting the real-life subjects right up close, having them do something insane with practical effects, and then CGI all around it, zoomed out. The CGI becomes window dressing for an already-impressive spectacle. That’s a film that will always be worth revisiting because of it. So I can be swayed on CGI. It’s just that tools will never be so good that you can use them unintelligently and still do good work. Legolas is a regrettable example of that.
Does it shame you not to have an interesting response? I thought you were here for conversation, my mistake I guess.
No it doesn’t. Because to me a reaction image and gif is EXCITING! and I felt like your comment slightly derailed the discussion already. Kind regards.
What discussion? Everyone else just flatly said no and I tried to say no in an interesting way. Whatever, boring ass.
