Sony Pictures to Use AI to Produce Movies and Shows in "More Efficient Ways"
hollywoodreporter.com
hollywoodreporter.com
Not yet. And the more niche the longer it will take for AI to be able to do it since you have less data for it.
Everything is derivative nowadays, AI or not. That's even how the schools themselves are structured.
We have ample evidence of technology increasing productivity.
There is a quality bar for Sony Pictures I assume.
My point is that we constantly invent technology to increase productivity. In this case, all I’m saying is that the cost to make a movie of the exact same quality should decrease with better tools.
If we gene edited humans to grow in the womb over 1 month then yes we can get a baby in 1 month.
Unfortunately, I don't see creative regional works growing as a result... Unless there is a way to channel these straight to the streaming services in some profitable way.
I am a director now.
So instead of buying a movie in a video format, we buy prompts.
A prompt can be something like “a horse enters the room”. But because it’s generated on the fly, the horse entering will walk around your kitchen island so it doesn’t bump into it.
So everyone’s experience will be slightly different but the story remains the same.
Right now, we are limited to a 2D screen for experiencing story telling. But I think technology is converging where we might completely change the way movies are presented.
Like you, I'm not sure I'd call it "like a movie"… but also not "like a video game" — it might be a new thing, part way between, that takes up time that would otherwise be spend on the aforementioned entertainment.
So we might not buy anything.
But I recently got a Vision Pro demo and watched some 3D scenes. I thought it was amazing and clearly more entertaining than a 2D.
Then I saw SORA and thought this is amazing too.
So combine them together and we might have a new way of experiencing stories. We don’t have to watch stories on a 2D screen. We can “experience” them in our own environment. Maybe share the same experience with friends and family who also have a VR headset on.
The status quo is that I can pay something like $10 a month to get access to hundreds of critically-acclaimed shows crafted be extremely skilled specialists that I can watch on my $300 55” 4K TV screen. The average American spends about 5 hours watching content (TV and digital) per day.
So let’s take the Vision Pro at $3500. We’re going to make the battery life 5x longer, we will also remove any straps or pieces of the device that touch your face since that messes up hair and makeup, so we can say now it’s a pair of glasses. And we’re going to reduce the cost from $3500 to $300 to match the price of a 55” TV from Costco. We will also add on-device AI processing so that this product can deliver on your vision. Which technology advancements in the next 10-20 years are going to make something like that remotely possible?
In the 15 years that we’ve had modern smartphones on the market when have they ever dropped in price by 10x or gained battery life by 5x? I can’t seem to buy a new iPhone for $60 that gets a week of battery life.
These sci-fi future visions are hilarious in how shortsighted they are, like when The Jetsons “predicted” living in the sky with flying cars. Of course the problem is that doing something like that has no benefits and a laundry list of drawbacks and impracticalities.
If your answer is yes, nothing I say will convince you.
If your answer is no, then something will replace content on 2D screens for story telling.
On your initial comment, there's already a {f|c}ringe group of LLM users who use [E]RP to participate in... scenarios (text, can generate pics using SD). It's hardly googlable, you may or may not want to dig into this, I warned.
Plenty of higher technology products fail when simple and cheap solutions compete with them.
If you want to tell me that 3D, depth, and space is main the draw, I could point to the various failures of 3D movies and television that were supposedly the technologies of the future.
I used to own a VR headset, a Valve Index. I loved it for what it was. Games on that system were immersive like nothing I had experienced before.
But there were numerous downsides.
Cost, not really solved by today’s products.
Eye fatigue, not solved even by the Vision Pro.
Group experience. Not really solved unless you’re a total hermit. Am I going to go to enjoy a SuperBowl party where all the attendees are wearing a VR headset? How does VR benefit the experience of watching shows with family/friends on the couch? Is it an upgrade to put on individual headsets and using virtual avatars? How do I cuddle and kiss my spouse while we watch a movie with a headsets on? How do I watch VR content at a bar or restaurant?
Multitasking: At best not an upgrade, at worst a downgrade. If I use a TV I can watch the show talk to my friend and order toilet paper on my smartphone that I already own at the same time. What upgrade am I getting there with VR?
Battery life. Not solved. Can I even get through a SuperBowl game on a headset? Is my SuperBowl party going to be a group of 30 people standing around the power outlet?
Nausea and VR acclimation. We could call this partially solved. A pretty high percentage of the population just can’t do VR physically.
Sweaty face, messed up hair, not solved. Period.
Compatibility with corrective vision: not really solved, every headset needs custom prescription lenses or needs space to fit glasses (this was uncomfortable in my Index, I was lucky to be able to wear contacts).
Visual fidelity: not solved, a standard TVs still beat the Vision Pro at pixels per FOV. 4K per eye includes all the periphery vision that doesn’t represent the main focus of your content.
Profit, not solved. The only companies making a profit on VR are companies like Valve that aren’t burning cash on R&D. How long will public companies like Meta and Apple tolerate wildly unprofitable product lines? which Apple product that sold under 500,000 units at launch is still with us today? Maybe the Mac Pro?
Let’s also not forget that high technology in the medium isn’t typically the driver of what is popular in content. The 5th highest grossing film in 2023 was a film shot on analog film technology that primarily used practical effects. The most watched show in 2023 was a cop procedural that has been on antenna television for over 20 years. The top selling video game of all time is Minecraft. Nobody really cares about being blown away by immersive visuals when they just want to watch deckhands on a chartered yacht reality show sleep around and gossip.
Do you think we'll have solved the problems you outlined in 5 years? 8 years?
I'd rather focus on the actual vision I laid out rather than the limitations we have today.
Microsoft and Google backed out of AR/VR and they were early adopters. Why would mega-tech companies with "unlimited money" like that back out of the market if it's so promising? Microsoft even seemed to have the B2B/industrial AR market cornered and they just dropped the whole thing. But then again, why should Microsoft spend $10 to make $10 on AR when they can spend $10 on Azure and make $20 selling basic compute services?
Meta is spending 5x more on R&D than it is making back in revenue from its Reality Labs division. This is 5 years after they launched their first consumer device, an eternity in technology. Compare that to the smartphone business where Apple was making wild profits from day one.
It's very important to note just how capital intensive VR/XR is to develop and deliver in a way that's truly reality-bending. The development time and effort is closer to a commercial airplane than most traditional technology products, like PCs that are designed with generally off-the-shelf parts and software. Seriously: Boeing spent $32 billion to develop the 787, Meta burns that on VR R&D in just 2 or 3 years and they aren't selling planes to make up for it.
So, the answer to your question is, absolutely not. These problems won't be resolved in 5-8 years. Too few have been resolved in the past 5. It's just all moving too slowly.
Let's just single out the battery life issue as an example of why these problems can't be resolved fast enough. Apple's silicon is the best product out there for VR/XR but there hasn't been a generational leap in their computing power since the M1 chip. Apple is constrained by the lithography that TSMC can produce so each new generation of processor is in that 10-20% range of improvement. Therefore we can expect their processing to be no better than about twice as fast as their current chips in around 5-8 years. Battery technology won't progress in density significantly between now and then, so what you're left with is the Vision Pro of 5 years from now possibly doubling the battery life or shrinking the battery so it doesn't need to be external anymore while delivering pretty comparable battery life. It's still going to be a situation where you need a big bulky ski goggle thing on your head messing up your hair and makeup.
Finally, we have to talk about sheer market size numbers and just how wildly far off VR/XR is from the kind of numbers you see with smartphones, TVs, etc. You need an almost impossible growth rate that just hasn't showed up even though we're on the third iteration of the market leader's product line, which is sold at a relatively affordable game console-like price.
According to Statista the total size of the XR market installed base is 50 million units as of 2024. They expect the total size to be 100 million by 2026.
Let's compare that to 5 years after the iPhone launched, 2012. Smartphone sales were 600 million in 2012 alone. The global television market is sized at about $250 billion (which is not that bad for devices that have an average selling price well under $1000 and last much longer than a smartphone or computer).
We've seen what it looks like when Apple launches a successful product. The Apple Watch and AirPods were flawed products in their first iteration but we all IMMEDIATELY saw them EVERYWHERE within a year or two of their launch. The Vision Pro is approaching 6 months past the launch date, and it's supposed to be the first immersive spatial computer where you can be productive anywhere anytime, the ultimate movie system for the airplane, the ultimate mobile workstation, and in all my time living in an affluent city with lots of people and going to the airport almost every month I've NEVER seen one in a single coffee shop, coworking space, on the street, on the plane, nothing. Which makes a ton of sense when we find out that Apple was barely able to sell 500,000 of them, and we hear rumors that they are delaying/canceling next year's model.
This feels like a future where it will be very easy to get it wrong and you end up with terrible content. An AI needs a personality to make interesting content or else it just grabs the bell-curve of what "personality" it has derived from its training. Then again, what is the right personality to make a great movie in Y genre? How can an AI possibly get this right when by definition it's an "average of learnt inputs"?
That’s what I envision.
I’m not sure about the “easy to get it wrong” part. I assume we will figure out how to control the experience such that it will always be a consistently good experience.
I’m not referring to an AI generating the script. The script is still crafted by humans - maybe with specific tools.
this is pretty much the only way we'll be able to keep entire colonies of people entertained during multi-generational journeys to distant stars
What studies like them don't understand is: once you go AI one day we'll go all-in on AI. Why should I bother watching some Sony Pictures movie a writer and direction came up with if I can just AI to create one for me in the first place with the parameters I like.
If it's all fake anyway...
So yes, in the future, creating 2D movies/shows will be trivial.