EG "make me a 120 minute Star Trek / Star Wars crossover". It should be more or less comparable to a big-budget studio film, although it doesn't have to pass a full Turing Test as long as it's pretty good. The AI doesn't have to be available to the public, as long as it's confirmed to exist.
1,000People are also trading
@VitorBosshard yet just a few comments below someone wrote:
> I highly suspect that with correct composition of agents (and lots of compute) it can be done even now.
the delusion is strong
@VitorBosshard Nice fall today, 35 to 27%. Did your comment caused an acceleration ? Maybe this market has been a bit forgotten, relatively to a year ago.
@Zardoru I said it a year or two ago and I still think it's true: this market has got very stodgy through KBC dynamics. People entrenching big limit orders and whales not bothering to correct the market because they know somebody is going to give them a better rate if they're patient.
@skibidist No I think that comment is correct, you mostly "just" need a very good Movie Director agent that's able to critique the scenes it generates. Plus gobs of compute. I think that'd get you a mediocre Hollywood-ish film with current tech.
Is that worth developing right now? No way.
(As other commenters below have mentioned,) I think it's easy to underestimate how much intelligence is involved in making a good film. In particular, I'm thinking about the amount of selection that goes into it.
Yes, there are dozens or hundreds of human intelligences directly doing the work. But also, each of those brains was shaped and selected by a massive network of other brains. In film school you show your short film to your classmates for feedback, and if most of them say "meh" then they've shaped you. A good screenplay is written after years of winnowing and experimentation; it's revised after being critiqued by a dozen readers; it's chosen from a stack by an agent and producer. And each of those selectors were good because they themselves were selected.
A good film is made by a vast superorganism. A gigantic hive mind.
Maybe it's easier for an AI because it doesn't need to replicate the process, only the result. But still. It requires smarts.
@Tsunombie to be fair, Italian Brainrot often doesn't make sense and LLMs know we'll like it nonetheless
@DavidBolin Technically this is evidence for "The first compelling Grok Imagine 30 min show is more than 3 months away". In order for it to be evidence for NO, you'd have to assume
1. he's overconfident about Grok Imagine enough to be off by more than two years on a low-digit-months estimate
2. no one is ahead of Grok Imagine on this front
Fair enough on 1 I guess, but I'm not convinced on 2.
New big-budget studio film
https://gizmodo.com/the-mummy-new-sequel-brendan-fraser-and-rachel-weisz-2000681497
@JimHays I think the idea is that movies are so dumb that AI should be able to do it easily.
@DavidBolin pretty much this.
When I think of big-budget studio films, I tend to imagine either my favorites, or academy award nominees.
I've mentioned The Mummy Returns in previous comments, for whatever reason it's the one that comes to mind when I think about all the other movies that are big-budget studio films. The bar for this market is high, but I don't think it needs to be higher than the next Mummy sequel.
