Mind and Iron: The out-of-nowhere startup that will turn us all into keyboard filmmakers
A company's Internet-shaking moment. Also, the future of meat.
Hi and welcome back to another fiery episode of Mind and Iron. I’m Steve Zeitchik, veteran of The Washington Post and Los Angeles Times and lead climatologist of this (very hot) news biodome.
If you’re new here, just a reminder that our platform aims to deliver all that’s happening in our AI and future worlds, with a refreshing lack of corporate agenda —we just focus on how this change will affect us humans. You can pledge support for our mission here.
A week away from the ol’ keyboard meant an abundance of news. Apple announced it will build various OpenAI features into its products, like a ChatGPT-equipped Siri. More on that as it rolls out. (Tl; dr, these will be incremental additions of AI into our lives, but increments is how leaps happen.)
More immediately, an AI startup called Luma Labs came basically out of nowhere last week to release the first mass-edition of an AI video program, Luma Dream Machine. Luma allows us to toss out a few words and spring a video into being — aka Big Tech’s persistent quest to turn us all into Steven Spielberg without any of that pesky work or talent. We took said machine out for a spin to let you know what it will do to our creativity and content landscape.
Also this week, FTC chair Lina Khan made some pointed comments about how her agency will go after Big Tech in ways that could shape AI development. Is she a hero or a zealot?
And did you follow all that wackiness about Nathan’s banning Joey Chestnut from its July 4 extravaganza because he has a deal with Impossible vegan hot dogs? A mustard-squirt of a story — but with spicy consequences. We talk to one of the country’s foremost alternative-meat gurus to explain the coming battle — as much cultural as scientific — over our diet.
First, the future-world quote of the week, on said eating habits.
"Someone should do a brand of Tofu and just call it Brofu, and the whole advertising campaign is just bodybuilders. Or a plant-based burger and you call it MAGA Burger and you get a Joe Rogan-type podcaster to promote it. I really think it would clean up.”
—Food futurist Paul Shapiro on how to create a world of meat without animals
Let's get to the messy business of building the future.
IronSupplement
Everything you do — and don’t — need to know in future-world this week
Dream Machine into action; the path of Khan; building a protein beast?
1. IN NOVEMBER 2022, OPENAI AMBUSHED THE TECH INDUSTRY BY RELEASING CHATGPT BEFORE MANY OBSERVERS THOUGHT THE COMPANY WOULD, OR SHOULD.
The move shocked competitors (which had been working on similar products), obsessed consumers and set off an arms race whose effects can still be felt as we commence the summer of 2024.
Well, a startup named Luma Labs just out-Neiled Neil.
At the end of last week, Luma swooped in and released its video tool, Luma Dream Machine, beating out rival OpenAI’s Sora, which the company demo-ed in February but has yet to release. For the first time ever, we regulars have a chance to tinker with an AI video tool — we have, in other words, the God-like power to speak an idea into being. Let there be light, shining down on dancing tigers in Hawaiian shirts.
Like Sora, Luma’s product allows for short videos to be created from text or image prompts — the company’s model scours a massive trove of videos and essentially synthesizes them into a custom-made video to match a user’s specs (of six or seven words). Luma employs a tech called Neural Radiance Field, or NeRF, to give it that authentic-but-still-gauzy feel that is fast becoming its own aesthetic.
Dream Machine essentially does for short videos what ChatGPT did for thank-you notes — allow for someone without any tools or skills to churn out something competent. Caveat: No AI-generated sound yet. But that weirdly can enhance the poetry, hyper-modern animation taking the form of a Golden Age of Hollywood silent film. The dream machine name is apt — everyone here is moving worldlessly, somnolently. And so, here we have it: everyone can create short videos.
Everyone of course can already create short videos simply by taking out their phone and shooting what’s happening around them, which is why we have 23 gajillion posts on TikTok and your Uncle Charlie sending you the play-by-play of his grandson’s coloring adventure. What I should say is anyone can create a fictional video, without all the messy business of casting, staging and shooting the thing. Did you ever dream of being a Hollywood producer who can toss off pitches and then sit back as your vision magically comes to life? Well, now you can live that dream (for 10-15 seconds).
I spent about that much time creating this video of bears doing the Blinding Lights dance — about how long it took me to think of such silliness and type in the request — and while the result hardly even looks like a dance, let alone anything The Weeknd might be proud of, it does feature some pretty cool animation and the earliest hints of what could be a funny video. (I was able to extend it a few seconds, but too much longer is a reach for now.)
So what does this mean? First, the requisite coolness allowance. The idea that we can literally type a film into existence is something to stop and appreciate. Professional text creations are one level; painterly images are another. But video, even sans sound, is a whole other level of complexity. Anything that bridges the gap between Joe Schlepper and a film student with years of experience — anything that has such democratizing power — is a notable development, regardless of the quality or utility. Memes allowed us all to be video curators. Lumas allow us all to be video creators.
That’s the to-be-sure part. But here’s where the concept gets muddled.
As we mused back during the Sora demo last winter, many of these videos are likely to be glorified screensavers — “there’s no reason to believe it can do anything more than merely supplement a good story. You can make animals pedal bikes on water or a lonely man move through the Tokyo snow. Great. Now give me a reason to watch it.”
Now that we can play with them, the truth of that statement really lands. One of the most prominent Luma videos thus far is people bringing the Distracted Boyfriend meme to video life. What does that mean? As little as you might imagine. Like, literally, all he does after he turns around is follow the other woman and whistle. Not groundbreaking.
Actually it’s the opposite of groundbreaking. Videoifying an image like this negates its power. Part of what made Distracted Boyfriend so influential was how static it was; we could Play-Doh the meme to make whatever point we wanted, whether on capitalism or punctuation. Not only was no one asking the question “what happens after he initially swivels his head,” seeing him carry on makes me forget about all the meaning I had projected on the image in the first place. Before it was a powerfully ambiguous tabula rasa. Now it’s just some bro chasing a girl up the street.
And that’s with memes that only ever existed as images. Some Luma tinkerers have taken to animating meme images that began as movie scenes — which, yes, gives us the ouroboros-like scenario of an image that started as a video now going back to being a video, only worse.
The best use case here might be for original ideas — a shard of zaniness that someone conceived on the spot (but didn’t have the credentials or financing to animate themselves). That could work nicely (if it can avoid potential copyright-infringement). Content like that doesn’t even need to go viral. In fact, I think the low barrier to entry, coupled with the fact that all these videos start to look similar after a while, will prevent them from going viral. More likely, these Lumas will be useful for a specific purpose and then just fade away.
Or, put another way, these little AI videos aren’t about creativity. We’ll have a million such things, all slightly different, like snowflakes, all melting quickly. No, Lumas and Soras won’t change Internet culture — they’ll change human conversation.
Right now many of us converse with memes or emojis. It’s passive — we find one that fits and grab it off the shelf. But now that we could create one anytime we like, it’s easy to imagine a customized video being our new go-to form of non-lexical communication. This is especially true if, in the coming VR and AR world, our daily interactions are going to involve a lot more video overlays. (Luma co-founder Amit Jain used to work on Apple Vision Pro, and one can image many integrations into such tech.)
I don’t want to say that the ability to speak videos into being could well change how we communicate but…I think it could well change how we communicate. For the past decade the ready availability of emojis and gifs, combined with the screens on which we all live, made human interaction much different than anything that came before. We would be flat-out naive to assume that the rise of a new class of finger-snap videos won’t hurl us into an even newer and stranger communication future.
It’s hard to know what shape this will all take, but a tool this easy, powerful and vivid— whether from Luna or Sora or other competitors now jumping into the game — can’t but infiltrate our daily interactions. Decry it if you like (and I…sort of do?), but this has long been the trajectory of our post-textual world, and spot videos will only accelerate the trend.
In this regard, the real change Luma and Sora bestow won’t be in how they elevate us into Steven Spielberg — it will be how they take everyday messaging further into the realm of David Lynch. Why use words when choreographed bears are right there?
2. FTC CHAIR LINA KHAN HAS BEEN ONE OF THE MOST ACTIVE REGULATORS ON BIG TECH THIS COUNTRY HAS EVER SEEN.
Under Khan, the commission has sued Amazon for alleged monopolistic practices in the retail space, sued Apple for alleged monopolistic practices in the smartphone market, sued Microsoft to prevent the takeover of Activision Blizzard, and sued Nvidia to prevent the takeover of a chip designer. They’ve not sued OpenAI yet, but there’s still time — Khan is at the helm until September, and the FTC opened an investigation into the company last year.
So given the new AI-based concerns about a few companies controlling so much data, exerting so much influence and generally taking over so much of our lives without any apparent research on, let alone regard for, the effects of such brain-outsourcing, how much can we count on Khan and the FTC to step in? She appeared last week at an event sponsored by the news platform TechCrunch and addressed exactly this point.
Here’s what she said when a moderator asked her a version of the aforesaid question.
“There is a lot of interest across D.C. and making sure that we are able to harness the opportunity and potential that these [AI] tools present while also making sure that these markets stay open and fair and competitive, rather than allowing certain types of bottlenecks or choke points to emerge in ways that could undermine that competition and that opportunity and that innovation.”
She continued, “I was out in Silicon Valley a few months ago, and it was really interesting to hear from those founders in particular about how right now there is a whole lot of opacity around who’s getting access to some of these key inputs, be it compute, be it the models, be it whether there is any guarantee that you’re not effectively feeding back proprietary information. And so I think, there’s a lot of excitement, but we’re also hearing some wariness that can emerge when you realize there’s a lot of power already concentrated, and then that power being concentrated could foreclose innovation and competition.”
It certainly does appear like the FTC is doing everything in its regulatory power to slow down the train. The agency just confirmed it is investigating a Microsoft partnership with an AI startup.
But:
Khan as FTC chair has to spend a lot of time talking in business-competition terms, a theme that has a relationship to, but is not fully aligned with, many of the deeper concerns about AI. (Like, if monopolies are reined in the companies will have less ability to perpetrate all those harms, but I don’t know that adding a few more competitors fully prevents them, either.)
Part of the problem is that all the ways AI can be used to dominate our lives is still unknown to the companies themselves, let alone to us; the harm a firm can do here goes well beyond the well-trod ground of consumer choice and high prices.
Also, the FTC tends to act when there’s an acquisition or other form of consolidation. But given the existing landscape, a tech giant doesn’t need to make a major acquisition to do harm; simply controlling so much of AI with their existing market power (eg, Apple and its devices) or ubiquity of its models (OpenAI and its popular GPTs) could do the trick.
Still, Khan has suggested she will use the regulatory tools in her power to limit these harms — and driven home that this is a fight that can and should spread beyond the FTC. I’m reminded of her exchange with Jon Stewart when she visited “The Daily Show” a couple months ago. Because her words were kind of reassuring, they’re worth recalling now.
The part-time Comedy Central host — who himself dodged the effects of Big Tech quasi- monopoly by jumping ship from Apple last year so he could report on AI — raised the crucial issue we and other concerned folk have been noting: If AI is going to govern so much of our lives, and if AI is controlled by so few companies, what does that mean for privacy, labor opportunities, truth-telling and, indeed, human existence itself?
Stewart: “It’s already being consolidated…they [Apple, Microsoft and Google] all buy AI startups and put them behind their paywall and they’re already having an arms race to see who will be either the monopoly, or this will be an oligopoly.”
Khan: “I think it just shows one of the dangers of what happens when you concentrate so much power and so much decision-making in a small number of companies (echoing, in the most 21st century of ways, indie-rock icon Thurston Moore’s famous “When youth culture gets monopolized by Big Business, what are the youth to do?”)
“The first thing we need to do is be clear-eyed that there’s no AI exemption from the laws of the books,” Khan told Stewart, adding how we’re still “reeling” from the consequences of regulators’ hands-off approach to Facebook et al in the 2000’s.
“Are you optimistic that we will be able to catch up to this in time before something truly catastrophic happens?” Stewart asked her.
“There’s no inevitable outcome here,” Khan answered. “We are the decision-makers, and so we need to use the policy tools and levers that we have to make sure that these technologies are proceeding on a trajectory that benefit Americans and we’re not subjected to all of the risks and harms.” Amen.
3. TO SAY WE NEED TO FIND ALTERNATE MEAT SOURCES IS A PRETTY BIG UNDERSTATEMENT — keeping ten billion cows, chickens, turkeys and pigs in the U.S. farm system annually just isn't sustainable, environmentally or in other ways.
To say that there are misconceptions about what those alternate sources could be is an even bigger understatement. In fact there are many ways to go about this, from the soy- and pea-based proteins that make up the Impossible and Beyond Meats on our grocery shelves to the meat-like "mycoproteins" being cultivated like beer to the lab-grown meats from animal cells that has been making major leaps in the last year with companies like Upside Foods and Good Meat. But there's often just as much confusion as clarity about the options.
One of the people clearing up the haze is Paul Shapiro. The Northern Californian runs "The Better Meat Company," which supplies meat-like ingredients based on those mycoproteins (for all you amateur Gregor Mendels out there) to food purveyors. He's also the author of the influential 2018 book "Clean Meat: How Growing Meat Without Animals Will Revolutionize Dinner and the World," which won plaudits from figures as diverse as Eric Schmidt and Kathleen Sebelius.
With Ron DeSantis recently banning lab-grown meat in Florida and competitive-eating icon Joey Chestnut going the other way and signing a deal with Impossible vegan hot dogs, it seemed like a good moment to talk with Shapiro about the future of the American diet. The conversation has been edited for brevity and clarity.
Mind and Iron: The number of animals in the U.S. farm system has gone up nearly 100 percent since the late 1980's even though the American population has only increased 35 percent. You've argued that this represents a real danger — ethically, environmentally and even just for our individual health. How can clean meats reverse the damage?
Paul Shapiro: There are some of us who see animal welfare as a concern with all this farming. But even for the many people who don’t, there are so many other reasons to reduce the very high number of animals in the system, especially environmental reasons. And to do that without giving up the taste of meat, which most Americans don’t want to give up. Which I don’t want to give up. That’s where clean meat comes in.
M&I: Buying traditional farm-raised meat is an ingrained habit for many American consumers. Why are you optimistic the availability of new products will change that?
PS: Simply, because we’ve done it before. Twenty years ago plant-based milk [almond, soy and oat] was almost unheard of. Now it's 15 percent of the market. Americans are willing to change their diet if it's tasty and affordable. Burger and fries are popular for a reason. Clean meat is getting cheaper and tastier. This will help people change their habits.
M&I: You've said, though, that market forces are not enough to bring about major change.
PS: I don’t think they are. Maybe this industry will grow without government support but it would happen a lot faster with it. Look at what happened with electric vehicles. The industry was moving very slowly until you saw states like California and the federal government drafting rules to stimulate it. I mean, Tesla would have been a goner without government support. And now EVs are ten percent of cars sold. Government has a big role to play here.
M&I: What would that support look like for alternative meat?
PS: We already have 150 clean meat companies in this country; they could be given incentives. Other companies could be given money to get off the ground. We're in a dry environment for capital, which makes it hard to build manufacturing facilities. There's a very clear case for governments to step in and build these facilities.
M&I: You believe it's a matter of global competitiveness too.
PS: Absolutely. China is zooming ahead on this; they made clean meat part of their five-year [food-supply] plan. We already lost out to them on solar panels and alternative energies. Do we want to do that again?
M&I: How much progress have we made on the science? Between the soy, peas and fungi — which you use at Better Meat Company — it seems like we’re extracting meat-like texture from an unprecedented array of sources.
PS: Plant-based meat has gotten quite good — we should be impressed with how good it’s gotten. The tools are there; it’s not like someone needs to invent a new calculus to keep improving. Standard food science can apply, microbiology and all the rest. That said, plant-based meat is still discernible from animal meat. Making it indistinguishable is what many food scientists shoot for. That’s the holy grail. But I also don’t know that it always has to reach that. Fifty years ago we didn’t eat nearly as much chicken in this country as we do now. We didn’t switch because it was a perfect mimicry of meat — we switched for price and health reasons.
M&I: Meat cultivated from actual animal cells — the work being done by Good Meat and Upside Foods — seems like it might actually achieve that mimicry. But the production process is just so expensive.
PS: I was in Safeway last night and two patties of Beyond Burgers were $6.99, which compared to conventional beef’s $5 a pound is more, but bearable. But if you’re looking at cultivated meat we’re talking a lot more — like 40x. Obviously they’ll get that down. But the question is how fast. The 2030’s is the earliest — and those are optimistic projections.
M&I: It’s always striking to me the people who can’t foresee a shift to other forms of meat — who say that current habits are too ingrained. The history of food argues very strongly against this kind of stasis, doesn’t it?
PS: I mean, for much of human history there was no such thing as cheese! We hadn’t figured out how to domesticate cows, let alone curdle milk. And now most Americans eat it every single day. Dietary habits change. And they change based on a host of factors. It would be naive to think that this won’t keep happening.
M&I: One variable that seems like it could slow that shift is how politicized this has become — this idea that ‘only Democrats’ eat alternative meat. Red-meat Republicans — literally. Or maybe confounding is the word.
PS: Nothing is more illustrative of this than when Cracker Barrel a few years ago put Impossible Sausage on the menu. They didn't take pork sausage off the menu — they just added an option. And they were excoriated by some conservatives. It's asinine. Both liberals and conservatives need protein; both liberals and conservatives like meat. We're all humans and all have the same biology. But somehow you can't eat a different kind of meat if you're conservative?
M&I: How can that be changed?
PS: I think we need to break the association of plant-based protein as somehow being ‘women's’ food. Millions of men throughout Asia eat tofu. Nobody calls them women. It's marketing, really. Someone should do a brand of Tofu and just call it Brofu, and the whole advertising campaign is just bodybuilders. Or a plant-based burger and you call it MAGA Burger and you get a Joe Rogan-type podcaster to promote it. I really think it would clean up.
M&I: OK, speaking of the culture wars, I have to ask — the whole Joey Chestnut thing. What do you think? It's perhaps telling —in a good way — that much of the coverage just focused on the Nathan’s-Impossible rivalry without dwelling on the alternative-meat aspect.
PS: When I was a kid, my father would take me to football games and ‘limit’ me to one hot dog — per quarter! Since men eat much more meat than women, it's quite nice to see the man who can eat the most meat in one sitting on the planet opt for a plant-based hot dog. Good for Joey. Perhaps he'll be a role model for kids today who eat like I did when I was 12.
M&I: So what would success look like for the alternative meat market in, say, 10 or 15 years, when those kids are adults?
PS: I’d go back to EV’s and the ten-percent threshold. If clean meats can do the same for animals — reduce the number in the farm system from ten billion to nine billion — I'd be OK with that. More than OK with that. I’d consider that a big win.
The Mind and Iron Totally Scientific Apocalypse Score
Every week we bring you the TSAS — the TOTALLY SCIENTIFIC APOCALYPSE SCORE (tm). It’s a barometer of the biggest future-world news of the week, from a sink-to-our-doom -5 or -6 to a life-is-great +5 or +6 the other way. Last year ended with a score of -21.5 — gulp. Can 2024 do better? The spring hasn’t been great. But summer is off to…an OK start.
MEMES ARE ABOUT TO GET REPLACED BY LUMAS: It’ll be interesting, at least. -1
LINA KHAN IS FIGHTING THE GOOD FIGHT ON AI: +2.5
A RECOGNITION IS GROWING THAT WE NEED ALTERNATE SOURCES OF MEAT: +2