the most advanced generalist AI models often outperform specialized models, even in the specific domains those specialized models were designed for. That means that if you want a model that can do a lot - reason over massive amounts of text, help you generate ideas, write in a non-robotic way — you want to use one of the three frontier models: GPT-4o, Gemini 1.5, or Claude 3 Opus.

Working with advanced models is more like working with a human being, a smart one that makes mistakes and has weird moods sometimes. Frontier models are more likely to do extraordinary things but are also more frustrating and often unnerving to use. Contrast this with Apple’s narrow focus on making AI get stuff done for you.

Every major AI company argues the technology will evolve further and has teased mysterious future additions to their systems. In contrast, what we are seeing from Apple is a clear and practical vision of how AI can help most users, without a lot of effort, today. In doing so, they are hiding much of the power, and quirks, of LLMs from their users. Having companies take many approaches to AI is likely to lead to faster adoption in the long term. And, as companies experiment, we will learn more about which sets of models are correct.

now the streaming gold rush—the era that made Dickinson—is over. In the spring of 2022, the Federal Reserve began raising interest rates after years of nearly free credit, and at roughly the same time, Wall Street began calling in the streamers’ bets. The stock prices of nearly all the major companies with streaming platforms took precipitous falls, and none have rebounded to their prior valuation.

Thanks to decades of deregulation and a gush of speculative cash that first hit the industry in the late Aughts, while prestige TV was climbing the rungs of the culture, massive entertainment and media corporations had been swallowing what few smaller companies remained, and financial firms had been infiltrating the business, moving to reduce risk and maximize efficiency at all costs, exhausting writers in evermore unstable conditions.

The new effective bosses of the industry—colossal conglomerates, asset-management companies, and private-equity firms—had not been simply pushing workers too hard and grabbing more than their fair share of the profits. They had been stripping value from the production system like copper pipes from a house—threatening the sustainability of the studios themselves. Today’s business side does not have a necessary vested interest in “the business”—in the health of what we think of as Hollywood, a place and system in which creativity is exchanged for capital. The union wins did not begin to address this fundamental problem.

To the new bosses, the quantity of money that studios had been spending on developing screenplays—many of which would never be made—was obvious fat to be cut, and in the late Aughts, executives increasingly began offering one-step deals, guaranteeing only one round of pay for one round of work. Writers, hoping to make it past Go, began doing much more labor—multiple steps of development—for what was ostensibly one step of the process. In separate interviews, Dana Stevens, writer of The Woman King, and Robin Swicord described the change using exactly the same words: “Free work was encoded.” So was safe material. In an effort to anticipate what a studio would green-light, writers incorporated feedback from producers and junior executives, constructing what became known as producer’s drafts. As Rodman explained it: “Your producer says to you, ‘I love your script. It’s a great first draft. But I know what the studio wants. This isn’t it. So I need you to just make this protagonist more likable, and blah, blah, blah.’ And you do it.”

By 2019, the major Hollywood agencies had been consolidated into an oligopoly of four companies that controlled more than 75 percent of WGA writers’ earnings. And in the 2010s, high finance reached the agencies: by 2014, private equity had acquired Creative Artists Agency and William Morris Endeavor, and the latter had purchased IMG. Meeting benchmarks legible to the new bosses—deals actually made, projects off the ground—pushed agents to function more like producers, and writers began hearing that their asking prices were too high.

Executives, meanwhile, increasingly believed that they’d found their best bet in “IP”: preexisting intellectual property—familiar stories, characters, and products—that could be milled for scripts. As an associate producer of a successful Aughts IP-driven franchise told me, IP is “sort of a hedge.” There’s some knowledge of the consumer’s interest, he said. “There’s a sort of dry run for the story.” Screenwriter Zack Stentz, who co-wrote the 2011 movies Thor and X-Men: First Class, told me, “It’s a way to take risk out of the equation as much as possible.”

Multiple writers I spoke with said that selecting preexisting characters and cinematic worlds gave executives a type of psychic edge, allowing them to claim a degree of creative credit. And as IP took over, the perceived authority of writers diminished. Julie Bush, a writer-producer for the Apple TV+ limited series Manhunt, told me, “Executives get to feel like the author of the work, even though they have a screenwriter, like me, basically create a story out of whole cloth.” At the same time, the biggest IP success story, the Marvel Cinematic Universe, by far the highest-earning franchise of all time, pioneered a production apparatus in which writers were often separated from the conception and creation of a movie’s overall story.

Joanna Robinson, co-author of the book MCU: The Reign of Marvel Studios, told me that the writers for WandaVision, a Marvel show for Disney+, had to craft almost the entirety of the series’ single season without knowing where their work was ultimately supposed to arrive: the ending remained undetermined, because executives had not yet decided what other stories they might spin off from the show.

The streaming ecosystem was built on a wager: high subscriber numbers would translate to large market shares, and eventually, profit. Under this strategy, an enormous amount of money could be spent on shows that might or might not work: more shows meant more opportunities to catch new subscribers. Producers and writers for streamers were able to put ratings aside, which at first seemed to be a luxury. Netflix paid writers large fees up front, and guaranteed that an entire season of a show would be produced. By the mid-2010s, the sheer quantity of series across the new platforms—what’s known as “Peak TV”—opened opportunities for unusually offbeat projects (see BoJack Horseman, a cartoon for adults about an equine has-been sitcom star), and substantially more shows created by women and writers of color. In 2009, across cable, broadcast, and streaming, 189 original scripted shows aired or released new episodes; in 2016, that number was 496. In 2022, it was 849.

supply soon overshot demand. For those who beat out the competition, the work became much less steady than it had been in the pre-streaming era. According to insiders, in the past, writers for a series had usually been employed for around eight months, crafting long seasons and staying on board through a show’s production. Junior writers often went to the sets where their shows were made and learned how to take a story from the page to the screen—how to talk to actors, how to stay within budget, how to take a studio’s notes—setting them up to become showrunners. Now, in an innovation called mini-rooms, reportedly first ventured by cable channels such as AMC and Starz, fewer writers were employed for each series and for much shorter periods—usually eight to ten weeks but as little as four.

Writers in the new mini-room system were often dismissed before their series went to production, which meant that they rarely got the opportunity to go to set and weren’t getting the skills they needed to advance. Showrunners were left responsible for all writing-related tasks when these rooms shut down. “It broke a lot of showrunners,” the A-list film and TV writer told me. “Physically, mentally, financially. It also ruined a lot of shows.”

The price of entry for working in Hollywood had been high for a long time: unpaid internships, low-paid assistant jobs. But now the path beyond the entry level was increasingly unclear. Jason Grote, who was a staff writer on Mad Men and who came to TV from playwriting, told me, “It became like a hobby for people, or something more like theater—you had your other day jobs or you had a trust fund.” Brenden Gallagher, a TV writer a decade in, said, “There are periods of time where I work at the Apple Store. I’ve worked doing data entry, I’ve worked doing research, I’ve worked doing copywriting.” Since he’d started in the business in 2014, in his mid-twenties, he’d never had more than eight months at a time when he didn’t need a source of income from outside the industry.

“There was this feeling,” the head of the midsize studio told me that day at Soho House, “during the last ten years or so, of, ‘Oh, we need to get more people of color in writers’ rooms.’ ” But what you get now, he said, is the black or Latino person who went to Harvard. “They’re getting the shot, but you don’t actually see a widening of the aperture to include people who grew up poor, maybe went to a state school or not even, and are just really talented. That has not happened at all.”

“The Sopranos does not exist without David Chase having worked in television for almost thirty years,” Blake Masters, a writer-producer and creator of the Showtime series Brotherhood, told me. “Because The Sopranos really could not be written by somebody unless they understood everything about television, and hated all of it.” Grote said much the same thing: “Prestige TV wasn’t new blood coming into Hollywood as much as it was a lot of veterans that were never able to tell these types of stories, who were suddenly able to cut through.”

The threshold for receiving the viewership-based streaming residuals is also incredibly high: a show must be viewed by at least 20 percent of a platform’s domestic subscribers “in the first 90 days of release, or in the first 90 days in any subsequent exhibition year.” As Bloomberg reported in November, fewer than 5 percent of the original shows that streamed on Netflix in 2022 would have met this benchmark. “I am not impressed,” the A-list writer told me in January. Entry-level TV staffing, where more and more writers are getting stuck, “is still a subsistence-level job,” he said. “It’s a job for rich kids.”

Brenden Gallagher, who echoed Conover’s belief that the union was well-positioned to gain more in 2026, put it this way: “My view is that there was a lot of wishful thinking about achieving this new middle class, based around, to paraphrase 30 Rock, making it 1997 again through science or magic. Will there be as big a working television-writer cohort that is making six figures a year consistently living in Los Angeles as there was from 1992 to 2021? No. That’s never going to come back.”

As for what types of TV and movies can get made by those who stick around, Kelvin Yu, creator and showrunner of the Disney+ series American Born Chinese, told me: “I think that there will be an industry move to the middle in terms of safer, four-quadrant TV.” (In L.A., a “four-quadrant” project is one that aims to appeal to all demographics.) “I think a lot of people,” he said, “who were disenfranchised or marginalized—their drink tickets are up.” Indeed, multiple writers and executives told me that following the strike, studio choices have skewed even more conservative than before. “It seems like buyers are much less adventurous,” one writer said. “Buyers are looking for Friends.”

The film and TV industry is now controlled by only four major companies, and it is shot through with incentives to devalue the actual production of film and television.

The entertainment and finance industries spend enormous sums lobbying both parties to maintain deregulation and prioritize the private sector. Writers will have to fight the studios again, but for more sweeping reforms. One change in particular has the potential to flip the power structure of the industry on its head: writers could demand to own complete copyright for the stories they create. They currently have something called “separated rights,” which allow a writer to use a script and its characters for limited purposes. But if they were to retain complete copyright, they would have vastly more leverage. Nearly every writer I spoke with seemed to believe that this would present a conflict with the way the union functions. This point is complicated and debatable, but Shawna Kidman and the legal expert Catherine Fisk—both preeminent scholars of copyright and media—told me that the greater challenge is Hollywood’s structure. The business is currently built around studio ownership. While Kidman found the idea of writer ownership infeasible, Fisk said it was possible, though it would be extremely difficult. Pushing for copyright would essentially mean going to war with the studios. But if things continue on their current path, writers may have to weigh such hazards against the prospect of the end of their profession. Or, they could leave it all behind.

Maybe one frame is through taking control of your own personal development and learning: after all “learning is the one thing your employer can’t take away from you”

Over the years we’ve seen the rise of bro-etry and cringe “thought leadership” and crying CEOs. When I scroll my feed I have to sidestep the clearly threadboi and #personalbrand engagement-farming posts and try and focus on the real content.

Networking is useful, but distasteful to many. Instead, participating in self-directed learning communities is networking

“Don’t become a marketing manager, become someone who knows how to run user research”

The “hard tech” in AI are the LLMs available for rent from OpenAI, Anthropic, Cohere, and others, or available as open source with Llama, Bloom, Mistral and others. The hard-tech is a level playing field; startups do not have an advantage over incumbents.

There can be differentiation in prompt engineering, problem break-down, use of vector databases, and more. However, this isn’t something where startups have an edge, such as being willing to take more risks or be more creative. At best, it is neutral; certainly not an advantage.

This doesn’t mean it’s impossible for a startup to succeed; surely many will. It means that you need a strategy that creates differentiation and distribution, even more quickly and dramatically than is normally required

Whether you’re training existing models, developing models from scratch, or simply testing theories, high-quality data is crucial. Incumbents have the data because they have the customers. They can immediately leverage customers’ data to train models and tune algorithms, so long as they maintain secrecy and privacy.

Intercom’s AI strategy is built on the foundation of hundreds of millions of customer interactions. This gives them an advantage over a newcomer developing a chatbot from scratch. Similarly, Google has an advantage in AI video because they own the entire YouTube library. GitHub has an advantage with Copilot because they trained their AI on their vast code repository (including changes, with human-written explanations of the changes).

While there will always be individuals preferring the startup environment, the allure of working on AI at an incumbent is equally strong for many, especially pure computer and data scientsts who, more than anything else, want to work on interesting AI projects. They get to work in the code, with a large budget, with all the data, with above-market compensation, and a built-in large customer base that will enjoy the fruits of their labor, all without having to do sales, marketing, tech support, accounting, raising money, or anything else that isn’t the pure joy of writing interesting code. This is heaven for many.

A chatbot is in the chatbot market, and an SEO tool is in the SEO market. Adding AI to those tools is obviously a good idea; indeed companies who fail to add AI will likely become irrelevant in the long run. Thus we see that “AI” is a new tool for developing within existing markets, not itself a new market (except for actual hard-tech AI companies).

AI is in the solution-space, not the problem-space, as we say in product management. The customer problem you’re solving is still the same as ever. The problem a chatbot is solving is the same as ever: Talk to customers 24/7 in any language. AI enables completely new solutions that none of us were imagining a few years ago; that’s what’s so exciting and truly transformative. However, the customer problems remain the same, even though the solutions are different

Companies will pay more for chatbots where the AI is excellent, more support contacts are deferred from reaching a human, more languages are supported, and more kinds of questions can be answered, so existing chatbot customers might pay more, which grows the market. Furthermore, some companies who previously (rightly) saw chatbots as a terrible customer experience, will change their mind with sufficiently good AI, and will enter the chatbot market, which again grows that market.

the right way to analyze this is not to say “the AI market is big and growing” but rather: “Here is how AI will transform this existing market.” And then: “Here’s how we fit into that growth.”

We still believe that there will be a separation between the “application layer” companies and foundation model providers, with model companies specializing in scale and research and application layer companies specializing in product and UI. In reality, that separation hasn’t cleanly happened yet. In fact, the most successful user-facing applications out of the gate have been vertically integrated.

We predicted that the best generative AI companies could generate a sustainable competitive advantage through a data flywheel: more usage → more data → better model → more usage. While this is still somewhat true, especially in domains with very specialized and hard-to-get data, the “data moats” are on shaky ground: the data that application companies generate does not create an insurmountable moat, and the next generations of foundation models may very well obliterate any data moats that startups generate. Rather, workflows and user networks seem to be creating more durable sources of competitive advantage.

Some of the best consumer companies have 60-65% DAU/MAU; WhatsApp’s is 85%. By contrast, generative AI apps have a median of 14% (with the notable exception of Character and the “AI companionship” category). This means that users are not finding enough value in Generative AI products to use them every day yet.

generative AI’s biggest problem is not finding use cases or demand or distribution, it is proving value. As our colleague David Cahn writes, “the $200B question is: What are you going to use all this infrastructure to do? How is it going to change people’s lives?”

both Netflix and Shein realised that you can make far more SKUs if you’re not constrained by physical inventory - the time slots on linear TV and the store rooms of physical retail.

If you don’t need thousands of physical stores, then you can turn over the product range much faster and reach new customers much more quickly - and so Shein is now bigger than H&M and on track to pass Inditex.

Of course, the fundamental TV question is ‘what’s your budget?’ There’s a circular relationship: a given budget means a given quality and quantity of content, which, combined with your CAC, means a given audience, which means a given level of revenue and a given budget. There is no network effect in TV, and going to Hollywood with the world’s best software and $5 will get you a latte.

YouTube doesn’t buy LA stuff from LA people - it runs a network, and the questions are Silicon Valley questions. YouTube, in both the network and the kinds of content, is a much bigger change to ‘TV’ than Netflix. It’s ‘video’, but it’s also ‘time spent’ and it competes with Netflix and TV but also with Instagram and TikTok (it does puzzle me that people focus on competition between Instagram and TikTok when the form overlaps at least as much with YouTube). And YouTube doesn’t really buy shows or buy users - it pays a revenue share.

Netflix can indeed make TV shows as well as any legacy TV company, but did Disney make software that’s as good as Netflix? It didn’t have to. It just had to make software that’s good enough, because ‘software’ questions are not the point of leverage. But I don’t see any media companies competing with YouTube or TikTok, where software is the point of leverage - at least, not recently.

What can you do with this thing? There’s a good chance that, whatever killer apps may emerge, they don’t need the entire complement of sensors and widgets to deliver a great experience. As that’s discovered, Apple will be able to open a second tier in this category and sell you a simplified model at a lower cost. Meanwhile, the more they manufacture the essentials—high density displays, for example—the higher their yields will become, the more their margins will increase. It takes time to perfect manufacturing processes and build up capacity. Vision Pro isn’t just about 2024’s model. It’s setting up the conditions for Apple to build the next five years of augmented reality wearable technology.

VR/AR doesn’t have to suck ass. It doesn’t have to give you motion sickness. It doesn’t have to use these awkward, stupid controllers you accidentally swap into the wrong hand. It doesn’t have to be fundamentally isolating. If this paradigm shift could have been kicked off by cheap shit, we’d be there already. May as well pursue the other end of the market.

what starts as clunky needn’t remain so. As the technology for augmented reality becomes more affordable, more lightweight, more energy efficient, more stylish, it will be more feasible for more people to use. In the bargain, we’ll get a display technology entirely unshackled from the constraints of a monitor stand. We’ll have much broader canvases subject to the flexibility of digital creativity, collaboration and expression. What this unlocks, we can’t say.

it is a seemingly inevitable consequence arising from the combination of the ease of changing how a platform allocates value, combined with the nature of a "two sided market," where a platform sits between buyers and sellers, holding each hostage to the other, raking off an ever-larger share of the value that passes between them.

Today, Marketplace sellers are handing 45%+ of the sale price to Amazon in junk fees. The company's $31b "advertising" program is really a payola scheme that pits sellers against each other, forcing them to bid on the chance to be at the top of your search.

Search Amazon for "cat beds" and the entire first screen is ads, including ads for products Amazon cloned from its own sellers, putting them out of business (third parties have to pay 45% in junk fees to Amazon, but Amazon doesn't charge itself these fees).

This is enshittification: surpluses are first directed to users; then, once they're locked in, surpluses go to suppliers; then once they're locked in, the surplus is handed to shareholders and the platform becomes a useless pile of shit.

This made publications truly dependent on Facebook – their readers no longer visited the publications' websites, they just tuned into them on Facebook. The publications were hostage to those readers, who were hostage to each other. Facebook stopped showing readers the articles publications ran, tuning The Algorithm to suppress posts from publications unless they paid to "boost" their articles to the readers who had explicitly subscribed to them and asked Facebook to put them in their feeds.

Today, Facebook is terminally enshittified, a terrible place to be whether you're a user, a media company, or an advertiser. It's a company that deliberately demolished a huge fraction of the publishers it relied on, defrauding them into a "pivot to video" based on false claims of the popularity of video among Facebook users. Companies threw billions into the pivot, but the viewers never materialized, and media outlets folded in droves:

These videos go into Tiktok users' ForYou feeds, which Tiktok misleadingly describes as being populated by videos "ranked by an algorithm that predicts your interests based on your behavior in the app." In reality, For You is only sometimes composed of videos that Tiktok thinks will add value to your experience – the rest of the time, it's full of videos that Tiktok has inserted in order to make creators think that Tiktok is a great place to reach an audience.

"Sources told Forbes that TikTok has often used heating to court influencers and brands, enticing them into partnerships by inflating their videos’ view count.

"Monetize" is a terrible word that tacitly admits that there is no such thing as an "Attention Economy." You can't use attention as a medium of exchange. You can't use it as a store of value. You can't use it as a unit of account. Attention is like cryptocurrency: a worthless token that is only valuable to the extent that you can trick or coerce someone into parting with "fiat" currency in exchange for it.

For Tiktok, handing out free teddy-bears by "heating" the videos posted by skeptical performers and media companies is a way to convert them to true believers, getting them to push all their chips into the middle of the table, abandoning their efforts to build audiences on other platforms (it helps that Tiktok's format is distinctive, making it hard to repurpose videos for Tiktok to circulate on rival platforms).

every time Tiktok shows you a video you asked to see, it loses a chance to show you a video it wants you to se

I just handed Twitter $8 for Twitter Blue, because the company has strongly implied that it will only show the things I post to the people who asked to see them if I pay ransom money.

Compuserve could have "monetized" its own version of Caller ID by making you pay $2.99 extra to see the "From:" line on email before you opened the message – charging you to know who was speaking before you started listening – but they didn't.

Useful idiots on the right were tricked into thinking that the risk of Twitter mismanagement was "woke shadowbanning," whereby the things you said wouldn't reach the people who asked to hear them because Twitter's deep state didn't like your opinions. The real risk, of course, is that the things you say won't reach the people who asked to hear them because Twitter can make more money by enshittifying their feeds and charging you ransom for the privilege to be included in them.

Individual product managers, executives, and activist shareholders all give preference to quick returns at the cost of sustainability, and are in a race to see who can eat their seed-corn first. Enshittification has only lasted for as long as it has because the internet has devolved into "five giant websites, each filled with screenshots of the other four"

policymakers should focus on freedom of exit – the right to leave a sinking platform while continuing to stay connected to the communities that you left behind, enjoying the media and apps you bought, and preserving the data you created

technological self-determination is at odds with the natural imperatives of tech businesses. They make more money when they take away our freedom – our freedom to speak, to leave, to connect.

even Tiktok's critics grudgingly admitted that no matter how surveillant and creepy it was, it was really good at guessing what you wanted to see. But Tiktok couldn't resist the temptation to show you the things it wants you to see, rather than what you want to see.

“We want the chronological feed back!” Instagram users scream into the void. “Here, have Reels and Shopping,” said Instagram’s CEO, on the hunt for new revenue streams.“We want freedom of speech!” tweet the denizens of Twitter. “But then our sponsored hashtags won’t be brand safe,” said Twitter’s CEO (whoever that is this week).

Coming back to the internet analogy, how did Google, Amazon etc ended up so successful? Metcalf’s law explains this. It states that as more users join the network, the value of the network increases thereby attracting even more users. The most important thing here was to make people join your network. The end goal was to build the largest network possible. Google did this with search, Amazon did this with retail, Facebook did this with social.

Collecting as much data as possible is important. But you don’t want just any data. The real competitive advantage lies in having high-quality proprietary data. Think about it this way, what does it take to build an AI system? It takes 1) data, which is the input that goes into the 2) AI models which are analogous to machines and lastly it requires energy to run these models i.e. 3) compute. Today, most AI models have become standardized and are widely available. And on the other hand, the cost of compute is rapidly trending to zero. Hence AI models and compute have become a commodity. The only thing that remains is data. But even data is widely available on the internet. Thus, a company can only have a true competitive advantage when it has access to high-quality proprietary data.

Recently, Chamath Palihapitiya gave an interview where he had this interesting analogy. He compared these large language models like GPT to refrigeration. He said “People that invented refrigeration, made some money. But most of the money was made by Coca-Cola who used refrigeration to build an empire. And so similarly, companies building these large models will make some money, but the Coca-Cola is yet to be built.” What he meant by this is that right now there are lot of companies crawling the open web to scrap the data. Once that is widely available like refrigeration, we will see companies and startups coming up with proprietary data building on top of it

DAOs are effectively owned and governed by people who hold a sufficient number of a DAO’s native token, which functions like a type of cryptocurrency. For example, $FWB is the native token of popular social DAO called Friends With Benefits, and people can buy, earn, or trade it.

Contributors will be able to use their DAO’s native tokens to vote on key decisions. You can get a glimpse into the kinds of decisions DAO members are already voting on at Snapshot, which is essentially a decentralized voting system. Having said this, existing voting mechanisms have been criticized by the likes of Vitalik Buterin, founder of Ethereum, the open-source blockchain that acts as a foundational layer for the majority of Web3 applications. So, this type of voting is likely to evolve over time.

On the flip side, this move remains the latest signal from the company that it’s done with being innovative and disruptive and has, as publicly traded companies usually do, shifted toward nickel-and-diming and turf protection as it attempts to fend off competitors.

Swelling trade tensions and the rise of “direct-to-consumer” platforms were bound to heighten the scrutiny on the import of mass media cultural products. But it’s also notable that the Marvel movies that did gain admittance in China were led by six heroes (The Avengers), five of whom were employed by the American military (with the sole outlier being an extraterrestrial) and all of whom were white. The current, rejected leads are more diverse in vocation, American allegiance, and ethnicity (among other attributes).

In 2017, Disney began a marketing integration with aerospace and defense giant Northrop Grumman encouraging those who use Google to research American defense contractor Stark Industries to join something like the real thing.

Avatar’s unprecedented achievements require us to examine not just its technological innovations, but also its narrative. The film’s “protagonist humans” are classic Western archetypes such as the taciturn soldier and the driven scientist. The villains are archetypes as well, but they are also particularly close to foreign caricatures of evil Americans: the tough-as-nails, violence-prone colonel and pillage-the-earth corporate executive. Furthermore, Avatar’s overarching message is one of collectivism, spiritualism, and alignment with nature. At the end of the movie, each of the Western heroes literally shed their individual identities (and white bodies) to become part of the cooperative aboriginal mind and save the day.