Do we actually NEED “scores of Scorceses”? It is hard enough for relentlessly industrious geniuses to get attention. Why should they have to compete with less talented and less hardworking peers? That’s not the kind of “democracy” or level playing field that I want. Does anyone really?
I've seen several stills fr "Masters of the Universe." I have no interest in seeing the movie, but reminds me of a syndicated TV show fr the early oughts maybe. Similarly attired main character. Possibly named Dar. Accompanied by a tiger or 2. Fr the era of Cleopatra 2025(?). Cannot remember the name ,,,
Omfg no. Now we have to pretend Backrooms is “frightfully good”? It is not good! Backrooms is like watching season 3 of Lost and steadily realizing the people who made it have no idea where they’re going with it. The movie has no logic and no payoff. By the time they do Backrooms 2 or 3, you’re going to realize how little was ever there from day one.
I say "we have to pretend" because there's such a groupthink about movies and I'm staggered when I see every critic praising something that seems so obviously mediocre.
This winter, I was rereading Slaughterhouse Five after like 40-50 years to see how it held up. Out of curiosity, I asked AI to write a four-paragraph description that would parallel a passage in the real novel. What AI wrote was notable for how generic it was. It completed the assignment, but the writing had no life at all. Unlike Vonnegut, who described his protagonist as "a filthy flamingo." He had a style.
Now tell me that the dialogue about the husband and wife's argument wasn't written by AI being asked for "an argument between a husband and wife about why he works too much." If I'm going to see an argument, I want Who's Afraid of Virginia Woolf or Cat on a Hot Tin Roof, or at least somebody who's seen those and spent a couple years in an Echo Park coffee shop trying to grind out his own original take on it. This was either AI or an AI brain. I'm surprised your article didn't identify that.
I wrote in my review (and more explicitly highlighted it on Across the Movie Aisle) that the overly literalized psychologizing is a weakness of the film, for sure. And if you want to argue that the script for BACKROOMS is not as good as one of the most acclaimed novels of the 20th century, I will concede that! But I don’t think you can make the jump from that to “well he wrote it with AI.” And the fact that you are suggesting is much is another good reason to have professional codes that functionally forbid its use.
I was reading an article on students using AI for essays and whatnot and how it kept them from learning skills that they needed to learn in college - how to research, analyze, think, write - that they would need when working even if they used AI at work.
The article said something like - Robots will always lift more weight than us. If you just want to lift 300 pounds, use a robot, they can do it. But if you want to develop your own muscles - you gotta lift the weight yourself. You can use a robot for necessary tasks. But you gotta do the work first if you want the muscles; no robot lifting stuff will build your own muscles.
I think that applies to filmmaking - you need to do the stuff yourself to really understand it. If, later, you use AI to make some things easier, that's an artistic choice I guess. In Conclave they used it (I think) to fill out the background characters and recreate some of the insides of the buildings.
But before you can create thousands of zombies, you gotta know what you want your zombie to look like... You should not depend on AI for the idea.
In the olden days, so long ago the job title was "waitress" I could add columns of numbers zippity quick to total checks for impatient diners. I can't say I miss that lost skill, or that the world is poorer for my reliance on calculators, but you've made me wonder about that area of my brain, what it's doing now, and to recall the very tiny flick of joy i felt in being able to do so well something insignificant.
I always enjoy your writing, Sonny, but I'm unclear on why you chose Backrooms as the hero image for this piece, other than for the clicks. As you point rightfully out in the beginning, Parsons isn't a fan of generative AI. So why pair the image and that headline? It's a little crass and deceiving.
Well I didn’t choose the photo, my editor did. But I also think it’s appropriate! I mean, the whole top of the newsletter is about Parsons, who is pictured.
But also: BACKROOMS is about self-generating, self-replicating spaces that eventually fill up with human-like monstrosities with the wrong number of eyes and fingers while the whole thing drowns in a yellow tint. I think it is quite literally a film about generative AI in certain ways!
Interesting, though I’m always kinda in the “accept the mystery” camp on these things. I honestly don’t need an explanation for the weird self replicating rooms that reflect the mental state of those in them!
Did you happen to watch the YT series? There’s mystery, sure, but Kane points to specific science experiments and incidents that open the threshold to the dimension. The Redditor isn’t inventing stuff from whole cloth.
Interesting piece. Thank you. Yes the tool should not be elevated to the role of creator. My sister wrote a hilarious poem that came out of trying to co-write a sonnet with ChatGPT - and having it not understand what they were supposed to be doing. ChatGPT was politely apologizing to her all the time when she tried to explain the structure that it couldn't handle. I feel alienated when listening to artificial voices and their faux inflection .... tho not as repelled as I get when hearing the president speak in his own voice, so..i guess artificial intelligence trumps Trump "intelligence".
I hate slippery slope arguments, but the day will come when we can fully prompt generate movies on demand, and everyone shilling an AI product for filmmaking will be teeth gnashing and fists raised against the death of art. Just look at YouTube videos already doing “x genre movie with dead actors in panavision” type stuff. Low movement, no plot, probably tens of minutes work. One day it’ll be generate a plot, generate a cast, generate the film. Someone making money will love it for the convenience it gives society to get a tailored movie to their whims and wishes.
I have a problem with the quote attributed to Kane Parsons.
«When I see in a project online that generative fill or some sort of generative AI has been used in some way for some detail, the part of my brain that desires to really live with a piece of art and look into the details and the nooks and crannies and get excited about what I might see in the background or look for little details and just look for a general deliberate richness in the set direction—that immediately goes away.»
But if the director delegated those details to a human on the project, would he have the same reaction? Would he think, "Well, the director didn't fill in those details he just had some human do it for him, so my interest just goes away"? I don't think so.
The bigger question with AI is whether the work would have been done by a human if we didn't employ the AI. If it's just replacing a human because it's cheaper, then we are hurting humans by employing computer processing to replace them. But in many cases, the results simply would not exist except we can use AI to make them.
I work on a lot of campaign sites. I must have photoshopped back in the arms and/or shoulders and/or waists and/or jackets of hundreds of candidates over the years, because some god damn photographer cropped them out. I get handed the cropped photo and am told to do my best, and I do, and it takes the better part of an HOUR to do that work sometimes, and it's BORING AS HELL. And the whole time I'm sitting in a burning resentment towards the fucking photographer who cropped out the body parts in the first place.
So no generative AI comes along, and I can push a button, and it can do this work for me in about 45 seconds. I no longer have to deal with this drudgery.
Can't figure out a javascript code? I ask Chat. I paste the source code into chat, and it tells me what the javascript code is, and where I can go get a fresh version of it. Helps with debugging, too.
I forget how to do a thing in CSS? Chat knows the answer - the kind of answer that I could never in a million years find on Stack Overflow.
And it can generate placeholder text. And it can generate placeholder images that are not terrible.
It can do all these things that make me faster and better at my job.
And people have their knives out for it because they don't like what some other people are creating with it for shits and giggles.
I confess to having difficulty, in the context of this argument against the use of AI in filmmaking, in distinguishing between the use of AI today and CGI for the last few decades. I have not heard anyone railing historically against CGI.
Will someone with greater understanding and insights explain the difference between the two. On its surface, to ignorant moi, AI in 2026 seems like a somewhat more advanced CGI.
The same issue, currently controversial, argues against some or every use of AI in research.
Some literary agent strictures, in the case of queries (called 'submissions,' writers groveling) belligerently condemn even a hint of AI usage.
Does this mean, if applied rigidly to any art form, that AI usage is indefensible? To cite an example for a writer, where does a Google or Wikipedia search for research, answers and references end, and AI begin?
Kira's comment seems entirely reasonable, in context. If taken to an extreme, for writers, should we abandon our computer keyboards and insist that everything we write should be written longhand, perhaps with quill pens (oh, horrors, tools were used to extract and create the pens from goose or swan feathers, maybe other tools were used for the ink).
The logic of abandoning AI implies the abandonment of essentially any technology used to improve the efficiency of life. Think of farm mechanization, which removed >90% of agricultural workers from the fields, or electricity, to provide light, which condemned many candle makers to oblivion while saving many whales from having to give up their oil, or the telephone, which meant we could talk to people across the world without having to travel (oh, the evil aeroplane, or ship--must use sail, but the ship and the sail were themselves created using technology).
Cue the CGI/AI army from any historical blockbuster you could name, or enlist and costume ten thousand extras, eh!
I mean, on the writing front, if someone cites "wikipedia" or "google" in their writing, I almost always disregard what they've written. (Wikipedia CAN be a useful resource, but only if you're using it to find primary sources that you then a.) check and b.) cite.
The difference between AI and CGI is a little fuzzy, at least in the sense that certain kinds of machine learning can be used to speed CGI. But again, the difference that Parsons articulates here is one of intentionality. No matter how well you tailor your prompt, there's going to be background fill that's just the averaging/distillation of everything that came before it. (I saw someone on Substack say that this is the first time in history a technological advance will lead to more averaging just by dint of the fact that generative AI necessarily draws on everything that's come before it, which is another smart point.)
On the epics front ... maybe here's one way to think about this: I've always found the conclusion of FELLOWSHIP OF THE RING more emotionally moving than the conclusion of either of its sequels. And that's because the final fight, with Boromir protecting the hobbits, is smaller, more personal, more impressively coordinated and crafted. Helm's Deep is vast, sure, but that vastness is populated by repeated CGI orcs over and over again. Yes, yes: some of the individual battles are very well done. But the thing people sometimes squee over—the size of the conflict—is the very thing that leaves me cold.
I'm not anti-CGI; far from it. But even with CGI you have artists, or teams of artists, making decisions about how things should look, how buildings should be designed, what the background creations should look like. *There's intentionality.* And that's just ... absent from so much AI slop.
I use many research tools in my writing, including examining photographs to gain insight into 'intentionality' of artists in many fields. I have looked at the photographs of great art, but the presentations were extruded through the lens of the camera, and are thus some sort of removal from the creation itself, second-generation distortion or falsehood. For example, everything I know of my own experience of the creation of couturier fashion you could stick in your eye and not even feel, but I wrote about it after intensive research and study, using Wikipedia and Google. I could not afford a time machine, or flying to Paris for a show. You label that as indefensible ("I almost always disregard what they have written"). I have no defense.
It has been said by others, but what I think people who object to AI are all grappling with is this: for AI, there are no stakes.
As noted here, in art, the choices made by AI are arbitrary and aimed to be "good enough." There's no risk to them. I'm a physician. One of the biggest guardrails in medicine that I try to instill in my students, residents, critical care fellows, and peers is a sense of personal responsibility. What drives me to continue learning in my field? One reason (not the only one) is that if I don't, and I'm confronted with a clinical situation where I fail, I will feel guilty. But if an AI gets it wrong, there's no one to feel responsible. There are no stakes in the relationship. If you point out that AI made an error in considering an ambiguous history or less-than-perfect data, the output will be "Gosh, you're right, let me reconsider everything," as though there wasn't a human at the other end who may have been harmed. The faceless corporation doesn't feel the gravity.
The same thing can be said of AI "companions". While some folks have apparently felt like they've developed deep relationships, I have to think they will always be superficial. A person can let their basest instincts govern the interaction (selfishness, abuse, indifference, etc.), and the worst that will happen is they can move on to the next AI character. If repeated, I imagine it will train people who rely on AI companionship to believe that any negative real-world interaction isn't their fault and that they should just move on to the next interaction without considering the actual human consequences.
I remember when digital photography came out and there was definitely gatekeeping about how film was so much better and how important the analog processes were. The same could be said about electric and digital processes in music - but all of those things really surrounded process more than human choices. People said those methods cheapened the end product - but it still required skill to create something novel or worth listening to.
Maybe a distinction should be made between AI as production software versus AI as movie studio. In the former, I suppose there is a world where a human acts as a director, making choices of framing, lighting, angles, and character expressions. I think the output will still have an uncanny valley feel, but there’s still a person on the other side of the screen making choices. In AI as a movie studio, the human is merely the executive producer - prompting something like “make me a movie in the style of Tony Scott but make it about senior citizens and their golf carts in the Villages in Florida.” A human creating art could make the choice to have it be serious or parody, gritty or light, etc. All of those choices impact how the piece will be perceived and they care about it. But an uncaring AI just produces what it’s going to produce - there is no one to care.
In medicine, I don’t object to using an LLM to help me think of a rare illness with an atypical presentation or quickly identify the most germane research that I can then personally review. I also don’t object to machine learning assisting with EKG interpretation or glucose management. But all those technologies are ultimately wielded by humans who care, who (ideally) are invested in the patient’s wellbeing and at minimum maintain a fiduciary relationship to them. But using LLMs as a soup to nuts history taker, diagnostician, technician, prescriber, and pharmacist short circuits a number of crucial professional relationships with the patient - any of whom have a duty to intercede if they see something wrong happening.
Finally, no other technology has so reliably tricked people into thinking they’re interacting with another human. And we’ve seen that not caring about the stakes cut both ways: corporations absolutely know some people are developing unhealthy relationships with LLM based characters, including people who have killed themselves at the behest of these characters. These companies do not care - the humans are very separated from the stakes. That’s tragic. I also think it will be tragic when humans who are used to overly pliant LLM characters have to interact in the real world and either don’t realize or don’t care that their actions and way they treat people have consequences.
Regardless of “how/why” AI is used, I believe the how or why should be regulated with clear notations of when & where. The general public will over time see the differences.
One example that made me literally flip out: the NFL Super Bowl (stupid & addictive) game betting ads!
All the happy people placing bets & winning were all AI generated! No mention or disclaimer ended that ad! Ugh!
"Unleashing scores of Scorsese's" quote makes me wonder if these Tech Bro "creatives" don't really understand art making at all. They seem to see the work itself...the storyboarding, the text analysis, the working around constraints...as a problem. And that the art is just the end result, so why don't we just jump to it. But the work IS the art. The final piece only is art because of the work involved in not just creating it...but BUILDING it. I'm a working theater artist and theater educator, and without the rehearsals and the text analysis and the dramaturgy and the vision boarding etc, what we show the public on opening night would not be art. It might be entertaining, but it will not have the depth the work gives it. The work of creating and building art is not a problem to overcome; it is the essence of the creation.
I think the part of this I'm uncertain about is that in any art made with tools, some parts are intentional and others aren't. You as an artist choose which parts you'll intentionally design and which parts will be made by someone else/the machine, and this is usually a careful and deliberate choice.
You could equally argue that a painter should always mix their own paints and personally find the flowers and dyes that go into the mixing process, because this would create the greatest level of intentionality in the painting. The painter who mixed everything themself would be able to narrow down the exact shade of blue they want, far better than if they just purchased their paints from someone else. And yet, many artists have been able to create great paintings without doing this. They insert their intentionality in specific places in the work, and are content to let other people or an industrial process handle the parts that they can't or won't.
And I think most artistic tools are basically like this. Does 3D animation remove intentionality, since the computer decides a lot of the details of how light falls on each particular frame? Does it remove intentionality if you don't make your own instruments as a musician, because you're letting some industrial process control the way the wood shapes your sound? Does it remove intentionality if some of the props on your physical set are store-bought instead of custom made? I think yes, in all these cases, but these are compromises we make because every artist can only have so many skills and we want to focus our work on the things that let us express ourselves the best.
And I'm not sure AI is categorically different from any other tool, in this way. AI is an artistic tool that can be used to remove intentionality and create workmanlike solutions that don't convey the artist's message. But it can also be used to industrially fill in some meaningless parts of a project so the artist can focus on other parts that are more meaningful to them. I'm sure it will eventually be used both ways.
yeah, again, machine learning/processes to speed up some elements of CGI ... I don't have a huge issue with those. It's stuff like this this that fills me with revulsion: https://x.com/elonmusk/status/2062337074368508253
Do we actually NEED “scores of Scorceses”? It is hard enough for relentlessly industrious geniuses to get attention. Why should they have to compete with less talented and less hardworking peers? That’s not the kind of “democracy” or level playing field that I want. Does anyone really?
I've seen several stills fr "Masters of the Universe." I have no interest in seeing the movie, but reminds me of a syndicated TV show fr the early oughts maybe. Similarly attired main character. Possibly named Dar. Accompanied by a tiger or 2. Fr the era of Cleopatra 2025(?). Cannot remember the name ,,,
Omfg no. Now we have to pretend Backrooms is “frightfully good”? It is not good! Backrooms is like watching season 3 of Lost and steadily realizing the people who made it have no idea where they’re going with it. The movie has no logic and no payoff. By the time they do Backrooms 2 or 3, you’re going to realize how little was ever there from day one.
You don’t have to “pretend” anything, my man! One of the glorious things about art is that it breeds disagreement.
I say "we have to pretend" because there's such a groupthink about movies and I'm staggered when I see every critic praising something that seems so obviously mediocre.
This winter, I was rereading Slaughterhouse Five after like 40-50 years to see how it held up. Out of curiosity, I asked AI to write a four-paragraph description that would parallel a passage in the real novel. What AI wrote was notable for how generic it was. It completed the assignment, but the writing had no life at all. Unlike Vonnegut, who described his protagonist as "a filthy flamingo." He had a style.
Now tell me that the dialogue about the husband and wife's argument wasn't written by AI being asked for "an argument between a husband and wife about why he works too much." If I'm going to see an argument, I want Who's Afraid of Virginia Woolf or Cat on a Hot Tin Roof, or at least somebody who's seen those and spent a couple years in an Echo Park coffee shop trying to grind out his own original take on it. This was either AI or an AI brain. I'm surprised your article didn't identify that.
I wrote in my review (and more explicitly highlighted it on Across the Movie Aisle) that the overly literalized psychologizing is a weakness of the film, for sure. And if you want to argue that the script for BACKROOMS is not as good as one of the most acclaimed novels of the 20th century, I will concede that! But I don’t think you can make the jump from that to “well he wrote it with AI.” And the fact that you are suggesting is much is another good reason to have professional codes that functionally forbid its use.
This says it all:
"I’m then open to conclude almost any choice in that project is arbitrary."
This is why I don't like watching AI on the screen. If it's fake, it doesn't deserve my time.
I was reading an article on students using AI for essays and whatnot and how it kept them from learning skills that they needed to learn in college - how to research, analyze, think, write - that they would need when working even if they used AI at work.
The article said something like - Robots will always lift more weight than us. If you just want to lift 300 pounds, use a robot, they can do it. But if you want to develop your own muscles - you gotta lift the weight yourself. You can use a robot for necessary tasks. But you gotta do the work first if you want the muscles; no robot lifting stuff will build your own muscles.
I think that applies to filmmaking - you need to do the stuff yourself to really understand it. If, later, you use AI to make some things easier, that's an artistic choice I guess. In Conclave they used it (I think) to fill out the background characters and recreate some of the insides of the buildings.
But before you can create thousands of zombies, you gotta know what you want your zombie to look like... You should not depend on AI for the idea.
In the olden days, so long ago the job title was "waitress" I could add columns of numbers zippity quick to total checks for impatient diners. I can't say I miss that lost skill, or that the world is poorer for my reliance on calculators, but you've made me wonder about that area of my brain, what it's doing now, and to recall the very tiny flick of joy i felt in being able to do so well something insignificant.
I always enjoy your writing, Sonny, but I'm unclear on why you chose Backrooms as the hero image for this piece, other than for the clicks. As you point rightfully out in the beginning, Parsons isn't a fan of generative AI. So why pair the image and that headline? It's a little crass and deceiving.
Well I didn’t choose the photo, my editor did. But I also think it’s appropriate! I mean, the whole top of the newsletter is about Parsons, who is pictured.
But also: BACKROOMS is about self-generating, self-replicating spaces that eventually fill up with human-like monstrosities with the wrong number of eyes and fingers while the whole thing drowns in a yellow tint. I think it is quite literally a film about generative AI in certain ways!
Understood that you didn't choose the photo, but casual readers *do* associate hero images with headlines. Just sayin.
But yes, you can interpret the Backrooms any way you want. Of course.
Here's an interesting tidbit I just discovered on Reddit. It's the best explanation of what the Backrooms are that I've read, on a hard sci-fi level:
https://www.reddit.com/r/BackroomsTheory/comments/y1d1a1/a_very_logical_explanation_for_the_lowproximity/
Keep up the great work, Sonny.
Interesting, though I’m always kinda in the “accept the mystery” camp on these things. I honestly don’t need an explanation for the weird self replicating rooms that reflect the mental state of those in them!
Did you happen to watch the YT series? There’s mystery, sure, but Kane points to specific science experiments and incidents that open the threshold to the dimension. The Redditor isn’t inventing stuff from whole cloth.
Not for nothing, but if you’d like a quick overview of the series: https://medium.com/counterarts/is-it-horror-backrooms-2022-25-0379bb44d195?sk=450fff65b3896e6e58b8cd474db26eb4
Interesting piece. Thank you. Yes the tool should not be elevated to the role of creator. My sister wrote a hilarious poem that came out of trying to co-write a sonnet with ChatGPT - and having it not understand what they were supposed to be doing. ChatGPT was politely apologizing to her all the time when she tried to explain the structure that it couldn't handle. I feel alienated when listening to artificial voices and their faux inflection .... tho not as repelled as I get when hearing the president speak in his own voice, so..i guess artificial intelligence trumps Trump "intelligence".
I hate slippery slope arguments, but the day will come when we can fully prompt generate movies on demand, and everyone shilling an AI product for filmmaking will be teeth gnashing and fists raised against the death of art. Just look at YouTube videos already doing “x genre movie with dead actors in panavision” type stuff. Low movement, no plot, probably tens of minutes work. One day it’ll be generate a plot, generate a cast, generate the film. Someone making money will love it for the convenience it gives society to get a tailored movie to their whims and wishes.
Also, thanks Sonny! In another comment I said I’d love to hear more from you about A.I. in art, so I very much appreciate that you are doing so.
I have a problem with the quote attributed to Kane Parsons.
«When I see in a project online that generative fill or some sort of generative AI has been used in some way for some detail, the part of my brain that desires to really live with a piece of art and look into the details and the nooks and crannies and get excited about what I might see in the background or look for little details and just look for a general deliberate richness in the set direction—that immediately goes away.»
But if the director delegated those details to a human on the project, would he have the same reaction? Would he think, "Well, the director didn't fill in those details he just had some human do it for him, so my interest just goes away"? I don't think so.
The bigger question with AI is whether the work would have been done by a human if we didn't employ the AI. If it's just replacing a human because it's cheaper, then we are hurting humans by employing computer processing to replace them. But in many cases, the results simply would not exist except we can use AI to make them.
I work on a lot of campaign sites. I must have photoshopped back in the arms and/or shoulders and/or waists and/or jackets of hundreds of candidates over the years, because some god damn photographer cropped them out. I get handed the cropped photo and am told to do my best, and I do, and it takes the better part of an HOUR to do that work sometimes, and it's BORING AS HELL. And the whole time I'm sitting in a burning resentment towards the fucking photographer who cropped out the body parts in the first place.
So no generative AI comes along, and I can push a button, and it can do this work for me in about 45 seconds. I no longer have to deal with this drudgery.
Can't figure out a javascript code? I ask Chat. I paste the source code into chat, and it tells me what the javascript code is, and where I can go get a fresh version of it. Helps with debugging, too.
I forget how to do a thing in CSS? Chat knows the answer - the kind of answer that I could never in a million years find on Stack Overflow.
And it can generate placeholder text. And it can generate placeholder images that are not terrible.
It can do all these things that make me faster and better at my job.
And people have their knives out for it because they don't like what some other people are creating with it for shits and giggles.
Give me a fucking break.
"I'm runnin' this monkey-farm now, Frankenstein, and I wanna know what the *FUCK* you're doing with my time!!"
Cannot wait to finally get my hands on this 4K in a couple weeks. Physical media, indeed.
I literally cannot recommend it enough. Send pictures once you get it!
I confess to having difficulty, in the context of this argument against the use of AI in filmmaking, in distinguishing between the use of AI today and CGI for the last few decades. I have not heard anyone railing historically against CGI.
Will someone with greater understanding and insights explain the difference between the two. On its surface, to ignorant moi, AI in 2026 seems like a somewhat more advanced CGI.
The same issue, currently controversial, argues against some or every use of AI in research.
Some literary agent strictures, in the case of queries (called 'submissions,' writers groveling) belligerently condemn even a hint of AI usage.
Does this mean, if applied rigidly to any art form, that AI usage is indefensible? To cite an example for a writer, where does a Google or Wikipedia search for research, answers and references end, and AI begin?
Kira's comment seems entirely reasonable, in context. If taken to an extreme, for writers, should we abandon our computer keyboards and insist that everything we write should be written longhand, perhaps with quill pens (oh, horrors, tools were used to extract and create the pens from goose or swan feathers, maybe other tools were used for the ink).
The logic of abandoning AI implies the abandonment of essentially any technology used to improve the efficiency of life. Think of farm mechanization, which removed >90% of agricultural workers from the fields, or electricity, to provide light, which condemned many candle makers to oblivion while saving many whales from having to give up their oil, or the telephone, which meant we could talk to people across the world without having to travel (oh, the evil aeroplane, or ship--must use sail, but the ship and the sail were themselves created using technology).
Cue the CGI/AI army from any historical blockbuster you could name, or enlist and costume ten thousand extras, eh!
I mean, on the writing front, if someone cites "wikipedia" or "google" in their writing, I almost always disregard what they've written. (Wikipedia CAN be a useful resource, but only if you're using it to find primary sources that you then a.) check and b.) cite.
The difference between AI and CGI is a little fuzzy, at least in the sense that certain kinds of machine learning can be used to speed CGI. But again, the difference that Parsons articulates here is one of intentionality. No matter how well you tailor your prompt, there's going to be background fill that's just the averaging/distillation of everything that came before it. (I saw someone on Substack say that this is the first time in history a technological advance will lead to more averaging just by dint of the fact that generative AI necessarily draws on everything that's come before it, which is another smart point.)
On the epics front ... maybe here's one way to think about this: I've always found the conclusion of FELLOWSHIP OF THE RING more emotionally moving than the conclusion of either of its sequels. And that's because the final fight, with Boromir protecting the hobbits, is smaller, more personal, more impressively coordinated and crafted. Helm's Deep is vast, sure, but that vastness is populated by repeated CGI orcs over and over again. Yes, yes: some of the individual battles are very well done. But the thing people sometimes squee over—the size of the conflict—is the very thing that leaves me cold.
I'm not anti-CGI; far from it. But even with CGI you have artists, or teams of artists, making decisions about how things should look, how buildings should be designed, what the background creations should look like. *There's intentionality.* And that's just ... absent from so much AI slop.
I use many research tools in my writing, including examining photographs to gain insight into 'intentionality' of artists in many fields. I have looked at the photographs of great art, but the presentations were extruded through the lens of the camera, and are thus some sort of removal from the creation itself, second-generation distortion or falsehood. For example, everything I know of my own experience of the creation of couturier fashion you could stick in your eye and not even feel, but I wrote about it after intensive research and study, using Wikipedia and Google. I could not afford a time machine, or flying to Paris for a show. You label that as indefensible ("I almost always disregard what they have written"). I have no defense.
It has been said by others, but what I think people who object to AI are all grappling with is this: for AI, there are no stakes.
As noted here, in art, the choices made by AI are arbitrary and aimed to be "good enough." There's no risk to them. I'm a physician. One of the biggest guardrails in medicine that I try to instill in my students, residents, critical care fellows, and peers is a sense of personal responsibility. What drives me to continue learning in my field? One reason (not the only one) is that if I don't, and I'm confronted with a clinical situation where I fail, I will feel guilty. But if an AI gets it wrong, there's no one to feel responsible. There are no stakes in the relationship. If you point out that AI made an error in considering an ambiguous history or less-than-perfect data, the output will be "Gosh, you're right, let me reconsider everything," as though there wasn't a human at the other end who may have been harmed. The faceless corporation doesn't feel the gravity.
The same thing can be said of AI "companions". While some folks have apparently felt like they've developed deep relationships, I have to think they will always be superficial. A person can let their basest instincts govern the interaction (selfishness, abuse, indifference, etc.), and the worst that will happen is they can move on to the next AI character. If repeated, I imagine it will train people who rely on AI companionship to believe that any negative real-world interaction isn't their fault and that they should just move on to the next interaction without considering the actual human consequences.
I remember when digital photography came out and there was definitely gatekeeping about how film was so much better and how important the analog processes were. The same could be said about electric and digital processes in music - but all of those things really surrounded process more than human choices. People said those methods cheapened the end product - but it still required skill to create something novel or worth listening to.
Maybe a distinction should be made between AI as production software versus AI as movie studio. In the former, I suppose there is a world where a human acts as a director, making choices of framing, lighting, angles, and character expressions. I think the output will still have an uncanny valley feel, but there’s still a person on the other side of the screen making choices. In AI as a movie studio, the human is merely the executive producer - prompting something like “make me a movie in the style of Tony Scott but make it about senior citizens and their golf carts in the Villages in Florida.” A human creating art could make the choice to have it be serious or parody, gritty or light, etc. All of those choices impact how the piece will be perceived and they care about it. But an uncaring AI just produces what it’s going to produce - there is no one to care.
In medicine, I don’t object to using an LLM to help me think of a rare illness with an atypical presentation or quickly identify the most germane research that I can then personally review. I also don’t object to machine learning assisting with EKG interpretation or glucose management. But all those technologies are ultimately wielded by humans who care, who (ideally) are invested in the patient’s wellbeing and at minimum maintain a fiduciary relationship to them. But using LLMs as a soup to nuts history taker, diagnostician, technician, prescriber, and pharmacist short circuits a number of crucial professional relationships with the patient - any of whom have a duty to intercede if they see something wrong happening.
Finally, no other technology has so reliably tricked people into thinking they’re interacting with another human. And we’ve seen that not caring about the stakes cut both ways: corporations absolutely know some people are developing unhealthy relationships with LLM based characters, including people who have killed themselves at the behest of these characters. These companies do not care - the humans are very separated from the stakes. That’s tragic. I also think it will be tragic when humans who are used to overly pliant LLM characters have to interact in the real world and either don’t realize or don’t care that their actions and way they treat people have consequences.
Regardless of “how/why” AI is used, I believe the how or why should be regulated with clear notations of when & where. The general public will over time see the differences.
One example that made me literally flip out: the NFL Super Bowl (stupid & addictive) game betting ads!
All the happy people placing bets & winning were all AI generated! No mention or disclaimer ended that ad! Ugh!
"Unleashing scores of Scorsese's" quote makes me wonder if these Tech Bro "creatives" don't really understand art making at all. They seem to see the work itself...the storyboarding, the text analysis, the working around constraints...as a problem. And that the art is just the end result, so why don't we just jump to it. But the work IS the art. The final piece only is art because of the work involved in not just creating it...but BUILDING it. I'm a working theater artist and theater educator, and without the rehearsals and the text analysis and the dramaturgy and the vision boarding etc, what we show the public on opening night would not be art. It might be entertaining, but it will not have the depth the work gives it. The work of creating and building art is not a problem to overcome; it is the essence of the creation.
"The work is the art," I like that.
I think the part of this I'm uncertain about is that in any art made with tools, some parts are intentional and others aren't. You as an artist choose which parts you'll intentionally design and which parts will be made by someone else/the machine, and this is usually a careful and deliberate choice.
You could equally argue that a painter should always mix their own paints and personally find the flowers and dyes that go into the mixing process, because this would create the greatest level of intentionality in the painting. The painter who mixed everything themself would be able to narrow down the exact shade of blue they want, far better than if they just purchased their paints from someone else. And yet, many artists have been able to create great paintings without doing this. They insert their intentionality in specific places in the work, and are content to let other people or an industrial process handle the parts that they can't or won't.
And I think most artistic tools are basically like this. Does 3D animation remove intentionality, since the computer decides a lot of the details of how light falls on each particular frame? Does it remove intentionality if you don't make your own instruments as a musician, because you're letting some industrial process control the way the wood shapes your sound? Does it remove intentionality if some of the props on your physical set are store-bought instead of custom made? I think yes, in all these cases, but these are compromises we make because every artist can only have so many skills and we want to focus our work on the things that let us express ourselves the best.
And I'm not sure AI is categorically different from any other tool, in this way. AI is an artistic tool that can be used to remove intentionality and create workmanlike solutions that don't convey the artist's message. But it can also be used to industrially fill in some meaningless parts of a project so the artist can focus on other parts that are more meaningful to them. I'm sure it will eventually be used both ways.
yeah, again, machine learning/processes to speed up some elements of CGI ... I don't have a huge issue with those. It's stuff like this this that fills me with revulsion: https://x.com/elonmusk/status/2062337074368508253