Convincing Human Thread Title Not Generated by AI

[domino harvey]

I mean, weren’t they more or less doing the young actress thing with the Angelina Jolie music video thirty years ago?

[paulm]

I mean, weren’t they more or less doing the young actress thing with the Angelina Jolie music video thirty years ago?

I remembered a Kristen Stewart one around 2012 too.

[beamish14]

I mean, weren’t they more or less doing the young actress thing with the Angelina Jolie music video thirty years ago?

I remembered a Kristen Stewart one around 2012 too.

They were minimizing their appearances in videos as far back as the mid-80’s. I absolutely adore this animated video for “The Harlem Shuffle” from Ralph Bakshi and John Kricfalusi that debuted at the 1986 Grammy Awards. The design of Keith Richards Puss in particular is just sublime

[hearthesilence]

I mean, weren’t they more or less doing the young actress thing with the Angelina Jolie music video thirty years ago?

I remembered a Kristen Stewart one around 2012 too.

Yeah, that was actually 2016 for Blue & Lonesome. It's the Stones, what do people expect? I wish the late period albums were better though.

[Lowry_Sam]

Tilly Norwood has a feature length film coming out.

[spectre]

I’ll believe it when I (don’t) see it:

https://variety.com/2026/film/global/ai ... 236802325/

“Misaligned” is currently in early development, with key collaborators currently being attached.

I wish people like the Guardian columnist in the link above would stop engaging with this desperate hype campaign, including by getting mad or making fun of it. On the small chance any such film ever does make it off the ground, my guess is that it’ll emerge in such diminished scope that any kind of box-office return will depend entirely on notoriety and hate-watching. So I don’t think the expression “if you ignore them, they’ll go away” has ever been more apt.

[hearthesilence]

After abandoning a $3.7 billion merger with rival Shutterstock in late June over a regulatory roadblock, Getty Images now plans to file for Chapter 11 bankruptcy and go private in partnership with Charles Koch’s investment firm.

It's relevant here because 1) half of its business dealt with creative stock media, and that side has crashed presumably because people have turned to AI to generate what they need, and 2) the other half dealt with actual news media - real life photos or footage that can be used for journalism - and a lot of the restructuring and future planning seems to hinge on the value of that side of the business, which is forecast to grow.

[Lowry_Sam]

More Perfect Union takes on AI in Hollywood animation.
(Merry The Christmas Witch vs Wildwood)

[domino harvey]

Between the release of the report speculating that we’re all going to die in two years from AI and Papa Elon providing users with a free anime sexbot, I thought it was a good time to check back in with AI. I don’t use it and I still can’t really understand why anyone would want a machine to think for them, but I am always curious about some of these grand visions being made about its ability to do specialized research and work.

I tested AI with a simple prompt that’s easy for me (and probably you) to answer but requires more specialized knowledge than you can get from models trawling the internet: If the Best Picture at the Oscars in the 50s had been expanded to ten spots instead of five, what would the other five films nominated each year likely have been? I asked AI to return its answers in order of likelihood and justify its reasoning. Before asking I produced my own list. Obviously there’s no right answer here, since it’s speculation, but there are def wrong answers and this thing is, as of yet, is not in danger of replacing anyone here

I won’t relay every iteration but for instance, in 1955, AI matches my two most likely nominations, East of Eden and Bad Day at Black Rock. These are gimmes, but still deceptively impressive. But then it makes a classic reasoning mistake: it clearly factors in current esteem, not contemporary attitudes. So it laughably posits that Night of the Hunter would be a likely nominee (possibly the least plausible answer possible, outside of a Martin and Lewis film). AI also hedged its bets by providing a list of five more “honorable mentions”… and yet, out of ten films suggested, none were Love Me or Leave Me, which I’d rank as extremely likely due to multiple factors (high profile nominations, a type of story Hollywood liked rewarding, the contemporary reception). Again, AI isn’t failing because it doesn’t agree with my predictions, it’s failing because it cannot imitate knowledge earned through watching these films, watching other films nominated in other years, recognizing trends that aren’t discussed on Reddit or awards forums, reading OOP books chronicling these years of awards and Hollywood that aren’t uploaded to Google books for AI to pull from, etc etc etc

I will say I used this same prompt two years ago on AI and its returns were virtually incoherent (for instance returning that for 1954, On the Waterfront would be nominated. Yeah, no shit, but that’s not what I’m asking), so the model is getting better. But I still have to question anyone claiming these services can replicate an actual expert in their field, unless being a film lover is now a lot more complicated than being a surgeon etc

I just tried this again for my annual single use of AI (with a different year, so it couldn’t just trawl my answer back to me) and AI completely nailed it. No notes, truly killed it. That just made me feel like that UCB sketch where they log roll to decide good vs evil: “Maybe we shouldn’t have rolled on that one”

[soundchaser]

It still doesn't quite pass the sniff test I set for it last year*, although it's closer. At least it's not hallucinating books that don't exist like the last time I ran this.

*List all the books containing Krazy Kat daily strips, along with publisher and ISBN (a scattered collection that isn't clearly documented, although Wikipedia's gotten more comprehensive).

[Never Cursed]

Between the release of the report speculating that we’re all going to die in two years from AI and Papa Elon providing users with a free anime sexbot, I thought it was a good time to check back in with AI. I don’t use it and I still can’t really understand why anyone would want a machine to think for them, but I am always curious about some of these grand visions being made about its ability to do specialized research and work.

I tested AI with a simple prompt that’s easy for me (and probably you) to answer but requires more specialized knowledge than you can get from models trawling the internet: If the Best Picture at the Oscars in the 50s had been expanded to ten spots instead of five, what would the other five films nominated each year likely have been? I asked AI to return its answers in order of likelihood and justify its reasoning. Before asking I produced my own list. Obviously there’s no right answer here, since it’s speculation, but there are def wrong answers and this thing is, as of yet, is not in danger of replacing anyone here

I won’t relay every iteration but for instance, in 1955, AI matches my two most likely nominations, East of Eden and Bad Day at Black Rock. These are gimmes, but still deceptively impressive. But then it makes a classic reasoning mistake: it clearly factors in current esteem, not contemporary attitudes. So it laughably posits that Night of the Hunter would be a likely nominee (possibly the least plausible answer possible, outside of a Martin and Lewis film). AI also hedged its bets by providing a list of five more “honorable mentions”… and yet, out of ten films suggested, none were Love Me or Leave Me, which I’d rank as extremely likely due to multiple factors (high profile nominations, a type of story Hollywood liked rewarding, the contemporary reception). Again, AI isn’t failing because it doesn’t agree with my predictions, it’s failing because it cannot imitate knowledge earned through watching these films, watching other films nominated in other years, recognizing trends that aren’t discussed on Reddit or awards forums, reading OOP books chronicling these years of awards and Hollywood that aren’t uploaded to Google books for AI to pull from, etc etc etc

I will say I used this same prompt two years ago on AI and its returns were virtually incoherent (for instance returning that for 1954, On the Waterfront would be nominated. Yeah, no shit, but that’s not what I’m asking), so the model is getting better. But I still have to question anyone claiming these services can replicate an actual expert in their field, unless being a film lover is now a lot more complicated than being a surgeon etc

I just tried this again for my annual single use of AI (with a different year, so it couldn’t just trawl my answer back to me) and AI completely nailed it. No notes, truly killed it. That just made me feel like that UCB sketch where they log roll to decide good vs evil: “Maybe we shouldn’t have rolled on that one”

Not trying to advocate for AI, but you can’t say this and not post it…

[domino harvey]

I already closed it out but I remember the order:

A Star is Born
Sabrina
Rear Window
the High and the Mighty
Executive Suite (and it wasn’t entirely confident in this choice so it offered Glenn Miller Story as an alternate)

I’d have gone with Barefoot Contessa over Executive Suite for the tenth spot, but it’s not an implausible choice (neither is the alternate)

[domino harvey]

I also asked it to discern my least favorite filmmaker and it did not disappoint



(you’d have to be a real old timer to get the other name it mentioned, who is not my least favorite but I can respect the deep cut)

Spoiler

Michael Mann

It still did quite poorly in picking my favorites, even with more specific prompting (identify my favorite French films), so there’s that at least

All of this to say that in no way does this make me want to use AI. If anything it accelerates the likelihood of me leaving the internet behind entirely (again?! We’ve seen this episode already) because Dead Internet Theory is here and now the machines really can start imitating the rest of us. Like those clowns in congress

[bdsweeney]

From what data is it working from to base its judgement? Your Letterboxd? Or streaming platform cookies? Do you know, Dom?

[HJackson]

I just tried this again for my annual single use of AI (with a different year, so it couldn’t just trawl my answer back to me) and AI completely nailed it. No notes, truly killed it. That just made me feel like that UCB sketch where they log roll to decide good vs evil: “Maybe we shouldn’t have rolled on that one”

It now "reasons" so that when you ask a question it will supply itself with additional context by asking and answering its own questions about what you asked before completing the expanded query, instead of just predicting the next most plausible words based on its training data. This makes the answers a lot more useful, while also making them more expensive to compute.

I have found that it still struggles with ambiguities of genre classification. One thing I asked AI repeatedly over the last couple of years was to identify a film noir I half remembered, in which a criminal is located by the police based on the knowledge that he enjoys the company of Asian women. I was not even sure if this film existed, or if I'd imagined it in some sort of vaguely racist fever dream. A few months ago it did identify the film correctly, but only after I asked about "a black and white American crime thriller" rather than "a film noir". If I query it as a film noir now, the AI invariably suggests The Crimson Kimono just based on a naive "film noir" + "Asian" = "The Crimson Kimono" calculation. In this sense it is still inferior to a knowledgeable human because I'm sure there are many people on this forum who could identify the film I was looking for based on my description of it as a film noir, and would supply the answer (perhaps with a caveat) regardless of whether or not they considered it a film noir, or a neo-noir, or a police procedural that no sane person should describe as any kind of a noir. The AI does not seem to be capable of that.

[domino harvey]

From what data is it working from to base its judgement? Your Letterboxd? Or streaming platform cookies? Do you know, Dom?

As HJackson noted, it actually provided reasoning. It used the awards the film was already nominated with (highlighting those with prestige AKA above the line) and cross referenced other awards (National Board of Review was one, can’t remember the rest) plus box office and past examples. I didn’t mention that it also floated one additional long shot nom (which it admitted) of White Christmas, based on the Oscars having a long history at this time of nominating the most successful film at the box office. I must admit I wouldn’t have it on my list but it was a surprising and on point possibility given the history of the category

I think the more important thing, which I guess I didn’t emphasize, is not even that it matched my best guesses, but its order of likelihood was also mine and that’s honestly what spooked me a bit

[knives]

It still doesn't quite pass the sniff test I set for it last year*, although it's closer. At least it's not hallucinating books that don't exist like the last time I ran this.

*List all the books containing Krazy Kat daily strips, along with publisher and ISBN (a scattered collection that isn't clearly documented, although Wikipedia's gotten more comprehensive).

I don’t think AI will ever be useful for that except when connected to a specific database (i.e. typing in a Wikipedia search bar). It’s main utility outside of mathematics seems to be in consolidating ideas such as when you need a set of documents with such and such information in such and such format.

[domino harvey]

Wow okay this thing still has a lot of work to do. I asked it to write a post in my style. One, nothing about this is in my style. Two, almost none of this is anything I’d ever say or think. Three, it sidelines Godard lol X 100. Four, when would I ever single out these names to represent my French cinema (Varda, Garrel, etc)? And five, a period at the end of every paragraph? My job is safe for now

The problem with French cinema, at least as it tends to be discussed in certain quarters, is that it has become almost impossible to separate the films themselves from the elaborate mythology surrounding them. We are forever being told about the Nouvelle Vague, about the radicalism of the jump cut, about the liberation of the camera from the tyranny of the studio, as though French cinema began in 1959 and everything that preceded it was merely a rehearsal for Godard smoking a cigarette and looking vaguely displeased with the conventions of narrative cinema.

And yet the more French cinema one watches, the less tenable this rather convenient historical shorthand becomes. Renoir understood the fluidity of human relationships long before anyone decided that a film could dispense with conventional dramatic architecture and still be considered an act of revolution. Ophuls could move a camera through a room with more emotional intelligence than most directors manage to muster across an entire career. Bresson, meanwhile, stripped cinema down to its essential components with such severity that one sometimes suspects he regarded the very idea of an actor as an unfortunate concession to commercial reality.

What interests me most about French cinema is not its supposed intellectual superiority, nor its frequently self-congratulatory relationship with the idea of artistic freedom, but its capacity for contradiction. It can be simultaneously austere and sensual, politically engaged and hopelessly bourgeois, formally adventurous and emotionally conservative. One can move from the moral labyrinths of Chabrol to the feverish romanticism of Garrel, from the immaculate formal constructions of Melville to the unruly pleasures of Pialat, without ever arriving at a single, coherent definition of what French cinema actually is.

Perhaps this is why the endless argument over the relative merits of the Nouvelle Vague has become so tiresome. There is a tendency to treat these filmmakers as though they constituted a unified movement with a common aesthetic program, when in reality their differences are often considerably more revealing than their similarities. Rivette's sprawling investigations into performance, conspiracy and the instability of reality have precious little to do with Truffaut's sentimental humanism, while Rohmer's apparently modest moral dramas can be more radical in their observations of human self-deception than Godard's most aggressively politicised experiments.

And then there is the matter of what gets overlooked. The French cinema of the 1930s and 1940s is too often reduced to a handful of canonical masterpieces, while the postwar period is treated as little more than a prelude to the arrival of the young Turks. Entire careers disappear beneath the weight of a few fashionable names. Directors who worked within commercial genres, who made melodramas, policiers, comedies and the occasional piece of unclassifiable lunacy, are relegated to the margins because their films do not conform neatly to the preferred narrative of cinematic modernism.

The irony, of course, is that French cinema's greatest achievement may be its refusal to remain within the boundaries imposed upon it by its admirers. Its history is messier, more commercial, more eccentric and considerably less respectable than the mythology suggests. There are masterpieces lurking in the most unexpected corners, and there are canonical works that become rather less impressive once one stops treating their reputations as an adequate substitute for actually watching them.

In the end, I would rather spend an evening with an obscure Delannoy melodrama or a disreputable little crime film from the margins of the French industry than participate in yet another ritual declaration of Godard's importance. Not because Godard isn't important, but because importance is a rather poor substitute for pleasure, and French cinema, at its best, offers an embarrassment of both.

Besides, any national cinema capable of accommodating Renoir, Becker, Bresson, Varda, Pialat, Tati and Jean Rollin without collapsing into an irreconcilable series of contradictions has surely earned the right to be approached with a little less reverence and considerably more curiosity.

Okay I’m done playing with this demon in a data center