James Cameron On AI in Hollywood
PLUS: Dyson's $3B haircare pivot, Taco Bell's Chief Food Innovation Officer.
Thanks for subscribing to SatPost (and thanks to all the readers that shared their glorious Dishoom dining stories after last week’s article).
Today, we’ll be talking about James Cameron, Avatar and how he thinks AI will change Hollywood.
Also this week:
Dyson’s $3B Haircare Pivot
Taco Bell’s Chief Food Innovation Officer
…them wild posts (including anon X accounts)
This issue of SatPost is brought to you by Bearly AI
Are you looking for a no-logging, encrypted and anonymized AI chat tool?
Then try the Bearly AI research app, which provides an easy-to-use UX for the newest models from OpenAI, Anthropic, xAI, DeepSeek, Gemini and more.
While some AI providers keep chat logs indefinitely even after a user deletes them...Bearly AI has a default no-logging policy and user chats are encrypted while requests to LLM providers are completely anonymized.
Try one month FREE of the Pro Plan using the code BEARLY1.
James Cameron on AI in Hollywood
The third film in James Cameron’s Avatar series is coming out on December 19th.
Based on the highest-grossing films of all time, let’s just say the first two films have done financially above-average (too lazy to adjust for inflation but you get the point):
Avatar (2009): $2.9B
Avengers: Endgame (2019): $2.8B
Avatar: The Way of Water (2022): $2.3B
Titanic (1997): $2.3B
Ne Zha 2: $2.2B
Still, Avatar might have the lowest cultural impact-to-box office ratio in Hollywood history. I’m not alone when I tell you I can’t utter a single quotable line from the series and might be able to identify 2 or 3 characters max. I know the blue people are Na’vi and there are floating rocks on the Earth-like moon called Pandora.
This is in stark contrast with the rest of Cameron’s filmography, which I can regularly quote out of context: The Terminator (1984), Aliens (1986), The Abyss (1989), Terminator 2 (1991), True Lies (1994), Titanic (1997).
When my house is out of toilet paper: “Game over man, GAME OVER!”
When I rattle off random facts about Vietnamese food: “My CPU is a neural net processor…a learning computer.”
When I think about the last time I made a good investment: “It’s been 84 years.”
Aliens and T2 are probably the two best action films ever while Titanic is my favourite summer blockbuster (I wrote more on the Leo, Kate and Celine classic here).
This is why I still plan on watching Avatar: Fire and Ash in the theatre, especially after a series of great Cameron interviews for his promotional tour.
A solid one is a two-part podcast with Matt Belloni on The Town (part 1, part 2), where the 71-year-old director shared some insights on how AI will change Hollywood.
Last year, Cameron joined the board of Stability AI and — since he’s a technological pioneer for film — is looking for ways to integrate generative AI into his work.
He’s aware of the AI downsides and tells Belloni that “I have lots of dualities in my mind that I’m very comfortable with.”
Cameron isn’t trying to cut his workforce. Rather, he sees the value of GenAI in speeding up production time for certain tasks, thus freeing people to work on more creative aspects of the job.
A typical Avatar film takes 1,000 VFX people and he intends to keep it that way:
[I want to work with] 1,000 people but for half the time. Then, we start the next one. So, nobody loses a job. Our cadence increases. Our throughput increases.
Look, I’m 71. I got a finite amount of time, maybe another 30-40 years at most.
More broadly, Cameron thinks the potential cost-savings of GenAI will be a positive for the industry by allowing the creation of new ideas that wouldn’t be financially viable otherwise:
[Look at] Dune or Wicked. They don’t have a lot in common as storytelling.
But they both have big sets and they cost a lot of money in VFX. […]
How many movies are going to get greenlit where you’re spending that amount of money on VFX that are not tried-and-true blue chip IP? Where does the young up-and-coming aspiring filmmaker — whose head is bursting with ideas like The Terminator or Aliens or The Abyss — get their foothold?
They don’t [right now].
[So, the VFX] has got to be cheaper.
He had a similar take with Meta’s CTO Andrew Bosworth in a podcast earlier this year:
So, I thought, I’ll join the board of a good competitive [AI] company…and Stable Diffusion was open source.
But my goal was not necessarily to make a shit pile of money. The goal was to understand the space.
To understand what’s on the minds of the developers. What are they targeting? What’s their development cycle? How much resources do you have to throw at it to create a new model that does a purpose-built thing?
My goal was to try to integrate [GenAI] into a VFX workflow.
It’s not just hypothetical. If we want to continue to see the kinds of movies that I’ve always loved and that I like to make…the [computer graphics] heavy films [have] to figure out how to cut the cost of in half.
Now, that’s not about laying off half the staff at a VFX company. That’s about doubling their speed to completion on a given shot. Cutting time in half. So your cadence is faster and your throughput cycle is faster.
Artists get to move on and do other cool things and then other cool things.
That’s my sort of vision for [GenAI]. I’ve been in the computer graphics game for 38 years. I’ve been there every step along the way.
We’ve almost plateaued evolutionary. It’s not revolutionary. And then this revolutionary technology comes and it’s coming at [computer graphics] from a completely different way of generating an image and they’re pretty incompatible [with traditional computer graphics] as it sits right now.
The trick is how do you increase that intersection of those two vendors.
Back to The Town podcast, Cameron doesn’t think AI will ever fully replace human actors and our individual-ness is the key going forward:
[GenAI is] trained on everything we ever valued artistically.
They don’t put the crap stuff into the training data. They put the stuff that’s been published. That’s online. The art. The performances. The movies.
Whether they’re doing it legally or ethically…to be determined.
Let’s assume that [GenAI] is almost de facto selection for that which the human eye and the human consciousness enjoys, right? That goes into the training data.
So, millions and millions of images and thousands of hours of performance. What you’re going to get is the average. It goes in a blender, right?
Then it precipitates out as a single unique new image, but it’s based on a sort of a generic feedstock. What [GenAI] can never do is create a unique lived experience reflected through the eyes a single artist. Whether that’s a single writer or single director or single actor.
[AI] won’t select for the quirkiness or the offbeat-ness.
I think what we celebrate is the uniqueness of our actors. Not their perfection, not their kind of glossy, Vogue-cover beauty. But their off-centerness.
I mean Sigourney Weaver or Cate Blanchett, they’re not like supermodels. They’re themselves, right? They’re not generic.
Celebrating actors is also the angle Cameron wants to use to regulate AI in Hollywood:
I think we as an industry need to be self-policing on [GenAI]. I don’t see government regulation as an answer. That’s a blunt instrument. They’re going to mess it up. […]
I think the guilds should play a big role. I think the Directors guild and the Actors Guild should play a big role. I mean, the Actors Guild certainly did it to the detriment of a lot of people, but they definitely drove a flag in the ground over this [with a strike in 2023].
It’s not a question of what we can legally do. Or even ethically what we should do. It’s a question of — I almost want to say morally — what we should do? What should we embrace? How should we celebrate ourselves as artists? How should we set artistic standard to celebrate human purpose?
This belief in the human-ness of the actor is why Cameron is very “annoyed” when people say Avatar is just “CGI” or “FX” or “AI”. It couldn’t be further from the truth:
For the making of Avatar 2 and Avatar 3, we did 18 months of motion capture.
[The actors] perform everything. Every breath is recorded, every bit of movement, every hand gesture…If you saw the characters underwater, the actors were underwater. If you saw them riding a creature, they were riding a water-powered jet. […]
I think we got burned a little bit by the advent of GenAI. At that point, people were conflating us with computer-made images.
Our images are performed by actors. They’re made by artists.
Sure, we use computers. You’re using a computer to read your notes right now.
It’s a tool. It’s not the creative force. The creative force is the human imagination.
For Cameron, motion-capture performances are actually the purest form of acting. Why? Because there is nothing the actor can hide. Every facial tick, body movement, breath and sound is captured (Belloni compares it to a super-intense form of Shakespearean stage acting).
If you’re a film or stage actor, please don’t yell at me for that take. Just passing the info along. Anyway, Cameron will highlight more of the acting in the newest Avatar):
I realized after the fact [that it] was a mistake to hide the man behind the curtain, so to speak. The man being Stephen Lang or Sam Worthington. All our digital artists — it takes over 1,000 people to make one of these movies — and they’re all artists. We have some engineers too, of course.
This time, I thought, “No, I got to reverse the polarity on this. I’ve got to show people what these actors are doing. The commitment they make. Their passion. How they put their heart and soul into the character as much as any live action production.”
With or without the actor emphasis, a third Avatar is about the surest thing ever to break $1B at the box office.
So, why such low cultural impact? A YouTube video by The Critical Drinker with 1.2m views tries to explain “The Avatar Paradox” and offers up a few reasons:
The characters and storyline aren’t very compelling
The 14-year gap between the first and second film took out narrative momentum.
While the first Avatar was a VFX breakthrough, the entire industry has caught up and it’s no longer as impressive.
Either way, check out this Insider video to see all the motion-capture VFX tech Avatar uses to show off the acting (the culmination of ~30 years of tinkering since Cameron first came up with the idea of Avatar in the mid-1990s).
Cameron’s emphasis on the acting itself reminds me a Pixar motto often repeated by Steve Jobs and John Lasseter in the studio’s early days: “No amount of technology can turn a bad story into a good story.”
Overall, his take on AI in Hollywood are very reasonable and balances the “dualities” in his mind.
But it’s worth emphasizing that Cameron is really only talking about blockbuster films and this is a tiny sliver of the market.
Starting from The Terminator, Cameron has only made 9 films. It was 12 long years between Titanic and the first Avatar (this was one of the most insane flexes btw; Titanic was the first movie to cross $1B box office and no one thought he could top it financially and he did).
Then, another 14 year until the second Avatar.
How many more blockbusters could he actually done with a higher cadence throughput? Three? Four? Five?
Doug Shapiro — who previously did Investor Relations at Time Warner and was Turner’s former Chief Strategy Officer — has a great series of articles on how GenAI will completely change the economics of Hollywood.
Here’s a notable excerpt from his latest piece “How Math Broke Media”:
When I talk to people about the effects of GenAI in Hollywood, they mostly think about it as a cheaper way to make TV shows or movies. That is a superficial (and skeuomorphic) way to approach it. The Hollywood studios, who are, in general, the most stuck in their ways, are experimenting with plugging AI tools into their existing workflows — like using it for concept art, previs, “subbing and dubbing,” or VFX.
GenAI-first studios shake their heads at this. They don’t have existing workflows, they are designing entirely new workflows and new job descriptions, built around GenAI. As a result, they are contemplating much larger savings in labor, time, and cost than the major studios. But I believe that even many of them are not thinking expansively enough. If you dig a little beneath the surface, you can see that their business models are quite similar to traditional studios—similar development processes and funding and monetization models.
The point of this example: even those positioning themselves at the cutting edge are arguably not thinking broadly enough about how much things might change.
I don’t think Cameron is wrong but the approach of adding GenAI to his VFX workflow probably doesn’t extrapolate far outside of blockbuster-type films meant for the theatre. He actually goes off on Netflix for being unfriendly to the theatre experience (and prefers David Ellison to own Warner Bros instead of the streamer):
[Netflix saying it would still do theatres if it acquired Warner is] sucker bait.
[They’ll] put it out for 10 days…to qualify for Academy Awards consideration. I think that’s fundamentally rotten at the core. A movie should be made as a movie for theatrical, and the Academy Awards to me mean nothing if they don’t mean theatrical.
I think [the Academy has] been co-opted and I think it’s horrific….[Netflix] should be allowed to compete if they put the movie out for a meaningful release in 2,000 theaters for a month.
In a best case scenario, GenAI could pause the decline of the theatre business and allow studios to take shots on new ideas and new directors. Instead of one OG making a $200m blockbuster, get four young bucks making the same type of extravaganza for $50m.
Honestly, not that bad of outcome based on the current trend line. Still a tiny percentage of total content that will be made with GenAI, though.
***
Finally, let me point you to another entertaining interview from Cameron’s promotional tour: an hour-long video with Vanity Fair that goes through each of his films.
Since we were on the topic of AI — and the fact that Cameron remains terrified of machine superintelligence — here are some nuggets on how he made The Terminator (1984) on a super tight budget of $4.3m:
approached the story as a low-budget horror slasher film
found areas in LA with the brightest street lights, to save money on lighting (”used car lots were great because they cast enough light from their floodlights onto the streets”)
cast Arnold (who hadn’t really broken through yet and was only paid $75K, which worked out to $1,293 per word)
The horror approach also worked because they could shoot a lot of the film guerilla style (no permits) in dark alleyways and parking garages. Smartly, Cameron layered on the sci-fi backstory to make it higher concept and laying groundwork for the sequel(s).
At the time, 30-year-old Cameron was living so frugally that his mom was sending him “2-for-1 Big Mac coupons” from McDonald’s because she didn’t think he was eating enough.
The film ultimately made $80m globally (20x-ish return), thus becoming a launchpad for Cameron and Arnold and quotable lines I could use in group chats (I repeat: “My CPU is a neural net processor…a learning computer.”)
Dyson’s $3B Haircare Pivot
Earlier this year, I wrote about “The Dyson Creep-Up”, the phenomenon when your home starts randomly gathering more and more Dyson products. It usually starts by stumbling into the vacuum section of Best Buy and — BOOM, a few months later — your living room has a Dyson fan and your wife’s bathroom cupboard has a Dyson hair dryer and hair straightener.
Well, Bloomberg recently wrote an article (“How Dyson Went From Fancy Vacuum Maker to Beauty Titan”) on the company and provided hard numbers to confirm my suspicions.
Since launching the Supersonic hair dryer in 2016, Dyson’s haircare category has grown to 30% of US sales (if we apply that percentage to the entire business, the haircare is about $3B of Dyson’s $9B+ in annual sales) and its tagline in the Sephora website goes very hard (“Engineering Beauty”).
The backstory is amazing and just classic James Dyson, who famously spent 15 years and made 5,127 design prototypes before releasing the first commercial Dyson vacuum in the early-1990s (he owns 100% of the business and is worth $15-20 Billy).
Dyson’s engineering knowledge — specifically of airflow and making tiny motors that spin very very fast — meant there were some obvious brand extension opportunities:
When Rob Smith [now Senior Principal Scientist at Dyson] interviewed with the company in 2012 for an engineering role, he assumed he’d be in vacuum cleaner development. Still, he told his interviewer that with the company’s knowledge of airflow “it would be really obvious if you’re going to go and do a hair dryer.”
He got the job and ended up being one of the first four engineers in what grew to be a 103-person team developing the Supersonic.
Apparently, Dyson himself was interested in haircare going back decades:
While Dyson, who stepped down as chairman in 2010, didn’t plan on making hair a core part of his business, he has had an interest in it at least since the 1960s, when he got his cut by Leonard of Mayfair, the celebrity hairstylist who gave Twiggy the crop that helped launch her modeling career.
I immediately Google’d “James Dyson hair” after reading that paragraph and, yooooooooooo, dude has flexed some incredible lettuce over the decades:
Dyson ended up spending 4 years and $71m to design the Supersonic Hair Dryer while slapping 100+ patents on it (eg. intelligent heat control, magnetic attachments).
The most impressive part is that this process included testing the device on 1,625km of hair. Probably would have been a complete failure if they only tested it on 1,624km of hair.
The Supersonic’s instant success — notably, it was much lighter than traditional brands (due to the smaller motor) — was followed by the Airwrap hair styler in 2018 and the Corrale hair straightener in 2020.
NO ONE IN THE WORLD CARES MORE ABOUT AIRFLOW THAN JAMES DYSON!
For real, watch the launch keynotes for the Supersonic and Corrale. Dude is obsessed.
Even at premium prices points ($400+), all of these products flew off the shelves and directly into my wife’s bathroom cupboard.
These were great examples of an engineering-first approach to beauty:
Airwrap Hair Styler: Dyson patented a special power cord that “rotates with the tool, solving the perennial tangling problem”.
Corrale Hair Straightener: Dyson designed a flexible copper plate that more evenly distributed heat across hair.
Having established itself in the $450B global beauty industry, Dyson is now making a play for other verticals including serums, sprays and creams.
In 2022, James Dyson personally poached a 20+ year vet from Estée Lauder (Kathleen Pierce) to become Dyson’s Global Head of Beauty. Her job is to combine Dyson’s innovate side with old school beauty know-how:
At traditional beauty companies, market data and expertise dictate product development, but Pierce says her priority is allowing engineers to apply design thinking to any hair problem and then getting out of their way.
“It’s about how many people can you put across a topic who have never thought about it before,” she says. “You never want to extinguish something.” Pierce’s own indoctrination into the company’s obsessive approach to detail was cycling through more than 80 bottle concepts for some of its hair lines, finally designing a pump that squirts precisely 0.22 milliliters (0.01 ounces) every time to prevent waste and stop users from applying too much.
It’s not clear if Dyson will find similar success in the chemical side of beauty products.
But James Dyson has always been more than happy to take wild bets. He whiffed on washing machines (“discontinued in 2005”) and electric vehicles (“abandoned in 2019”) but, somehow, “now operates as one of the UK’s largest farmers” (and will use some of these ingredients in the beauty products).
Meanwhile, Dyson’s latest bet is size-able: the company has spent $667m on beauty R&D in the past 4 years (most of this is done at Dyson’s 50 acre campus in the “ancient rural English town of Malmesbury”, which is thousands of miles from the company’s HQ in Singapore).
And Bloomberg’s review of the Dyson website found that 45% of the 78 available items are haircare products. I am too scared to do a similar review of my home but I’m guessing it is in the same ballpark. Ugh, well played James Dyson.
Taco Bell’s Chief Food Innovation Officer
The Wall Street Journal has a great piece on Taco Bell’s Chief Food Innovation Officer Elizabeth Matthews and the opening goes so hard:
A team at Taco Bell has been working, for nearly a decade, to invent a taco shell made entirely of cheese.
Its market research shows consumers are into the idea, and past inventions, like taco shells made from Doritos chips, were home runs that helped drive sales at the chain of around 7,500 U.S. restaurants.
It is on its fourth cheese supplier, and still is trying to figure out how to keep the cheese from disintegrating while the shell waits for fillings like ground beef and yet-more cheese.
In the fast food world, Matthews is considered the food innovation GOAT including some absolute classics:
Baja Blast: In 2004, Taco Bell was noticing that customers weren’t buying many drinks. The brand wanted to create a special product that could only be bought in its stores. Matthews went to PepsiCo HQ “for research and sampled sugary, caffeinated sodas for six hours straight” (she said of the experience: “By two o’clock in the afternoon, they put me in a conference room. They’re like, you’re going to need to just take a little nap.”) Eventually, she created the Mountain Dew-inspired drink and it’s been a top-selling mainstay.
Doritos Locos Tacos: Launched in 2012, the Doritos taco shells drove “hundreds of millions in sales in the first year and helping Taco Bell reverse a sales slump.”
Those two wild inventions were spin-offs from PepsiCo. This was a natural match because Taco Bell — which was founded in 1961 — was acquired by PepsiCo in 1978 before being spun off as the company now known as Yum! Brands (along with KFC and Pizza Hut) in 1997.
Matthews joined Taco Bell as a junior product developer two years after the spin-off.
She previously worked as a product developer making pizzas for Minnesota-based food company Schwan’s and operations for Garden Fresh Restaurant Corp.
One of her most important jobs at Taco Bell is to dream up limited time offers (LTOs), which McDonald’s does with the McRib or Starbucks does with Pumpkin Spice Latte.
LTOs are the best way to make sure customers return to see the latest demented invention and “the holy grail is a viral sensation online”.
A failed product was adding butter to rice. Meanwhile, recent LTOs include Nacho Fries, Crispy Chicken Nuggets, Pepper Steak Quesadilla and something you really shouldn’t put into your body the Baja Blast Pie.
Matthews oversees 100 people on the food innovation team and their customer research process is lit:
Each year, Taco Bell tests hundreds of ideas in the Test Kitchen. Promising ones go before what the company calls “sensory panels” of consumers to taste test. About 40 ideas advance to restaurant testing. Matthews is ultimately the one who decides when to greenlight ideas—or pull the plug.
Aside from food items, Matthews helped Taco Bell launch its Cantina sub-brand that offers alcohol including spiked Twisted Freezes (since 2015, its grown to 50 locations).
When they crack that all-cheese shell, you know your boy will be making a visit to Taco Bell…and possibly re-creating this meme:
Links and Memes
Some other links for your weekend consumption:
The Anti-AI Bet: On the Invest Like The Best podcast, Ari Emanuel explains why his portfolio of live assets (UFC/WWE, tennis, Frieze art fairs, car auctions) is the Anti-AI bet because we are social animals and the value of curated IRL events is going up as people get more free time from AI-powered productivity.
“The hidden cause of cultural stagnation”: Chris Dalla Riva says copyright incentivizes IP management over creativity. Hence, the endless reboots and sequels. The solution, shorten the period of copyright and expand “compulsory licenses”, which “allow a third party to use a protected work without the owner’s consent, provided royalties are paid to the owner”. So, you can use people’s work without asking but also have to pay an automatic royalty.
Why Movies Just Don’t Feel “Real” Anymore: A great 30-minute deep dive on how the use of CGI and digital tools end up making films look and feel fake (Avatar is actually an exception here; it looks real, but the stories just aren’t memorable).
Piers Morgan interviews Novak Djokovic: Man, Novak is the god damn GOAT. He won’t say it, though. Because he’s a student of tennis and says he’s building on legacy of others. It’s a great interview on his process and legacy.
Nvidia’s Bad Week: Since becoming the first company to hit $5T, Nvidia has fallen by $700B. Its clearly feeling the heat from short-seller Michael Burry (Nvidia sent a memo to investors saying that it is “not Enron”, which is literally the first thing they teach you in Investors Relations not to do) and Google’s in-house TPU chips, which are very competitive and Meta may start using at scale (Nvidia sent a very insecure post about Google, which is the first thing they teach you in Corporate Social Media not to do). Nvidia is still far and away the AI chip leader, but when you become the most valuable company in the world during a bubbly period when investors are looking for any reason to dump…few slip-ups ain’t good.
“we tried the worst pizza topping combination known to man”…an incredible play-by-play of ordering a custom Papa John’s Pizza with toppings that include pineapple, jalapeño, mushroom, extra anchovies, BBQ sauce and NO cheese (I just puked in my mouth).
…and them wild posts:
Finally, we need to talk about X rolling out a new feature that revealed the location of every account. It’s the worst-kept secret in social media land that many popular accounts on X talking about American politics are run by foreigners.
There’s always been state-level foreign psy-ops on Twitter because information warfare obviously. But since X started offering creator revenue in July 2023…ugh…let’s just take Charlie Munger’s quote “show the me incentive and I’ll show you the outcome.”
Someone with 100k followers that created a lot of engagement bait might score $500 a month or something in the ballpark. That’s not a lot for most North Americans but huge in the developing world.
Sure enough, since the location unveil feature on X, a ton of political accounts (on both the left and right) have turned out to be run from Pakistan, Nigeria, India, Bangladesh and around Eastern Europe etc.
Pirate Wires breaks it all down here.
These slop farms obviously work because there is demand for this type of content…so, just be mindful before spending 6 hours arguing with TrueHonestPatriot76 about the best Thanksgiving side.
Why advice Bearly AI when it does not work at all? Every chat disappears, there is no history, there is no usage data but pricing says that I'm limited by $20. At least that trial was free.
EDIT: And they did not even send me an email about subscription start. Do they hope to charge people at least once?