As those who follow me know, this isn’t a typical post for me—there’s certainly more important things going on. But unfortunately it impacts me personally, so I’ve been thinking about it a lot, and think there are larger ramifications that extend beyond mere entertainment.

This summer, Instagram updated its policies, and in particular is now cracking down on so-called “duplicate” content—photos and videos reposted directly on the feed (vs. shared on stories or using their recently launched “repost” feature). Many of my favorite accounts on Instagram are these exact type of meme accounts: ones that curate roundups of various types of content, typically crediting the original creators in both the text body and through tagging. While it’s impossible for me to get my hands on Instagram’s exact numbers, my assumption is that this ecosystem is, on balance, a win-win—helping both the meme page grow its audience and pointing that audience toward the original creators.

While my primary accounts, @jessedamiani and @urgentfuturespod, are not meme accounts exactly, I’ve participated in this fashion as well. It’s a way for me to exercise my curatorial chops in a lightweight way, drawing together various forms of content that include actual news, memes, silly videos, and lefty political content in 5-20 slide carousels. I view this act of curation as “transformative” in itself; even though I’m not changing the original pieces of content, by situating them together in one post, I’m attempting to capture a mood, share information, and implicitly contextualize current events.

Sometimes, these have done quite well, garnering tens and hundreds of thousands of views. Here’s a recent example:

Last week, I received a notification that my reach had been restricted to only my followers—none of my content would now show up on Reels, the Explore page, or anywhere else non-followers might otherwise find my content. This would persist, the notification explained, until I removed the content in question. It includes essentially everything in the meme batch linked above and several others.

While I’ve had isolated instances of content being auto-flagged because of rights issues (typically related to the music used in the video), I’ve never experienced something this overt. Across my feed, other accounts I follow, like @postp0stpost and abnormalize.being, to name just a couple, are bemoaning receiving similar notifications. To get a sense of what an account like POSTPOSTPOST is trying to accomplish, check out my conversation with founder Al Hassan Elwan on the Urgent Futures Podcast:

This might seem like a frivolous thing—and I can’t completely dismiss that critique—but on the other hand, there are stakes to this. If you’re like me, you rely on accounts like this for key cultural, social, and political commentary. They are essentially alternative “news” sources—if we broaden our understanding to understand news outside of traditional media. While these accounts don’t often produce original content (“OC”), their orientation to information and curatorial efforts serves as helpful filter through which to process news and information.

In some cases, this is simply entertainment. In others, these roundups operate at multiple levels, both bringing joy and making implicit arguments about life, the universe, and everything. A great example in this regard is ’s superb @nature_is_not_metal, which is not only a great place to encounter examples of human-animal and animal-animal cooperation, but an account that foregrounds necessary ideas like animism, non-dualism, etc.

Likewise, ’s @taylorlorenz3.0 is not only a great account to visit for a laugh, but a destination for news and hot takes in digital culture. Of course, there are loads more accounts like this, each that provide multiple layers of value along their respective niches.

Why This Matters

Meta, Instagram’s parent company, has explained that this policy change is designed to reduce “spam” and “low-quality behavior,” and incentivize “originality.” I can understand this core motivation, but they’re taking it way too far—and in the process making a huge mistake.

Share

I’ll grant that there are plenty of accounts that are just posting individual viral videos as their own, without crediting the original creators, and that this might detract from engagement that should be going to the original creators. While I’m not altogether convinced even by that argument, I can at least see how Instagram believes that removing this behavior will benefit the platform.

But it’s throwing the baby out with the bathwater, decapitating a thriving meme community that, for many of us, is among the primary reasons (if not the primary reason) that we spend time on Instagram in the first place.

Where this is perhaps most urgent has to do with issues of justice and equity. These meme accounts are routinely the first places where I encounter recordings of atrocities committed in Gaza and Sudan, critical climate news, and curated commentary on U.S. politics (especially from voices often excluded from mainstream sources)—to say nothing of the creators I’ve discovered and followed through exposure to their talents in these venues, folks from across the spectrum of comedy, music, dance, art, and otherwise.

The more cynical part of me wonders if this newfound crackdown by Instagram—after years of benefiting from this type of content—is an effort to cut off these communities in a bid to further appeal to Donald Trump and the MAGA movement.

Certainly there are right-wing memers who will suffer under this policy shift, but I believe the impacts will be felt much more acutely among progressive accounts. Why? Well, in the contemporary Western media ecosystem, the spectrum of what we consider mainstream has vanishingly little representation of leftist perspectives (yes, even on MSNBC) and the firsthand content that inherently links to these political views. A quick example of this in practice: compare the coverage of Russia’s invasion of Ukraine (which certainly deserved the coverage it got, and ultimately didn’t even receive enough of it!) and Israel’s of Gaza (which was equally as deserving, but received a fraction of the coverage). For many of us, Instagram was the primary vehicle by which we could get a real sense of what was happening on the ground in Gaza.

Right-leaning views (and recently, even right-wing extremism), on the other hand, are increasingly platformed, both in “mainstream media” and a bevy of more fringe outlets. So even if there are an equivalent number of meme accounts across the political spectrum who will be hit by this change equally, the role that left-leaning accounts play in the news ecosystem is proportionally greater for their respective communities.

For this reason, as well as the simple engagement/entertainment quotient, this move feels like a massive mistake for Instagram. We no longer live in the age where social media participation is a given. Look no further than X, which has seen a large-scale exodus since Elon Musk assumed ownership, driving tens of millions to Bluesky. It’s not just users deleting accounts, it’s also the “quiet quitting” aspect—a route that I’ve taken, alongside many of my peers, in which we leave our accounts intact so that they can’t be usurped, but we rarely or never use them.

Will there be a mass exodus from Instagram? Probably not. But I can imagine that this move will decrease the amount of time many of us actually want to spend on the platform. Given Instagram’s business model, decreased time-on-platform—even if we all keep our accounts—hits their bottom line. Just as network effects scale “up” when social media platforms are acquiring users—the more people who join, the more people feel like they have to be there—so too do they work “down”; the less time people spend on the platform, the less time that people will collectively spend on it, as it won’t be where they can find the types of content they prefer to engage with.

Of course, I could certainly be wrong here. I thought Netflix’s choice to crack down on shared accounts would cause more flight from the streamer than it ultimately did. From a financial standpoint, that’s proven effective. Likewise, maybe Instagram’s hypothesis about original content will prove out—frustrating people in the short term, but fostering enough ongoing interest that users don’t ultimately spend more time on other platforms. It seems like a risky bet to me, and worse: an unnecessary one. Time will tell; all I know is I’m bummed that my own experience of the platform has just been massively downgraded, and moreover that we’re experiencing yet another attack on the ability for key left-leaning content and information to circulate.

To that end, most of my work is already here on Substack, and I’m now more incentivized than ever to start bringing my meme roundups to Notes, so if you’re not already subscribed to Reality Studies or following my personal account, here’s your chance:

It’s always worth checking yourself when you feel a sense of doom. As Panic! At the Disco reminds us, “It’s much better to face these kinds of things with a sense of poise and rationality.” Are your concerns founded? Are they priorities (because boy-o, we don’t lack for things to feel concern about these days). So I’ve been self-evaluating for a bit as my thoughts coalesced.

It started a few weeks ago, as my timeline was increasingly dotted with videos of folks deepfaking OpenAI chief Sam Altman into various scenarios, running the gamut of banal to silly to alarming. Then I started seeing friends and colleagues deepfaking their likenesses into popular IP. A whole host of other Sora-2 trends and viral videos sprung up. The latest is cats with guns.

This is the scenario that I—learning from security researchers much smarter than me—have been concerned about for years, dating back to my reporting on deepfakes and mis-/disinformation in the 2010s and early 2020s. Video generators have been getting exponentially more capable over the past few years, so I knew this moment was going to come at some point. Google’s release of Veo 2 was the first time it felt like it was happening, and Sora 2 has sealed it for me.

Subscribe now

I don’t mean to condemn anyone who used Sora 2 this way, or who is curious to. It’s a perfect hack of evolutionary psychology—of course we’d want to try out a tool like this. Moreover, I continue to believe that even the most critical AI scholars have a responsibility to understand how tools work (i.e., ”know your enemy”), and some folks will learn best through direct experimentation. I discuss this in a bit more depth on a recent episode of Urgent Futures with media scholar Danny Snelson (for that and so many other reasons it’s worth a listen/viewing!):

Still, it has filled me with dread. And then a few days ago it got worse. OpenAI launched Atlas, it’s “agentic browser,” a chromium browser that has ChatGPT built in natively to do tasks for you. An example of what this looks like in practice: you find a recipe you like and then ask ChatGPT in the sidebar to assemble your Instacart order for you. Now apply this to every facet of your web experience and you have a sense of the allure of such a tool.

Like “deja vu all over again,” I witnessed folks in my life barrel headfirst into what feels like genuine nightmare scenario, reporting on their experiences with Atlas with breathless glee. Many began grandly proclaiming that this was The Future™.

Let’s leave aside all the hand-wringing about the hypothetical rise of “artificial general intelligence,” or AGI, at some point in the future; the developments listed above clock in as two critical risks AI poses to us right now.

Why Atlas and Sora 2 Scare Me

So let’s get into it. Why is this such a big deal? For different but related reasons, I’m concerned about the development of both of these products. How should we think through all of this and navigate it? And more fundamentally, why is OpenAI shipping products that hold questionable benefit but pose clear risks?

Though I think the scariest possibilities lie with Atlas, Sora 2’s impact to video is more visceral, so we’ll start there.

Sora 2 and the Deluge

You might guess that my biggest concern about realistic synthetic media is the usual one around its enablement of disinformation in which public figures are portrayed saying/doing things they didn’t actually do. This is indeed an evolving dimension of daily life we’ll now contend with—and that’s not ideal—but on balance these fears are overstated relative to what I think the thornier problems are.

Yes, you’re probably going to end up with egg on your face for rage-reposting a public figure saying or doing something they didn’t in the months and years to come, but in the end these are relatively easy to debunk. The damage is basically the whack-a-mole we all now have to play in coming to the truth. That’s annoying, yes—deeply so—but relatively low on the existential risk register.

One of my biggest concerns is actually the inverse of this scenario: how bad actors will weaponize the existence of generative video models to dismiss human-recorded footage. For example, that Sora is emerging while Republicans continue to use every tool in their toolkit to block the release of the Epstein Files is literarily problematic. Say Congress finally does succeed in forcing the files into the public eye, and it includes more imagery such as the Epstein birthday book—now anyone implicated will be able to claim that these are faked assets as part of a “witch hunt.” It doesn’t matter that it won’t be true, or that many will immediately know it’s not true—it will create a storm of diverging opinions and arguments that distract from justice being served (especially in a context in which the cards are stacked against those trying to hold the perpetrators to account).

Which brings us to another big concern—and I recognize it might sound banal at first—the volume of “bullshit” (in the Frankfutrian sense) that will soon exist in our digital lives. Both our respective individual attention spans and our ability to collectively organize around any one issue have already been fractured and atomized by algorithmic social media, which incentivizes engagement over civility. The rise of fast, cheap, plausible synthetic video is an accelerant to these tendencies, with some unique attributes that further exacerbate the situation. Everyone having the ability to, within seconds, produce plausible videos of anything simply means we’re going to spend ever more of our time swimming through the slop ocean, less being exposed to those trying to use their channels for positive, sensible, or just ends.

It goes further. Until now, there has been a fragile consensus that if a given video looked “real” enough, it probably was. There’s a sense in which this has always been a lie—we’ve been proverbially “photoshopping” photography since the birth of the medium (to the best of our knowledge, that honor goes to Hippolyte Bayard in 1839; see the video below)—but that collective agreement meant that we didn’t feel the need to treat each and every video as detective work, sleuthing out whether or not it was captured in the physical world. Sure, we did this in isolated scenarios (e.g., the moon landing video, JFK assassination, and more recently the Charlie Kirk assassination videos), but we weren’t forced to implement it as a broad practice.

No longer. Now even the most mundane video will necessarily involve a skeptical filter in which we have to ask and determine the veracity and creator of a given video—what Stephen Marche calls the “Big Blur.” I encountered this with a colleague the other day, who explained to me that they’d seen a video of Trump promising student loan debt forgiveness. “Was that real or AI?” they asked. I said that I didn’t know—I hadn’t seen the video—but it doesn’t track with any of Trump’s positions, so it seemed unlikely to be real.

This is where the forensic aspect comes into stark relief; we’re so keyed up to think about politicians being deepfaked to say damning things that “get them in trouble,” but what about throwaway “good” things that distract us from the real actions public figures are taking? It’s this register I worry about—the ways that existing filter bubbles, biases, hopes, or otherwise can be confirmed and enclosed, conditioning users to retreat into cozy imaginary realities where, as Kurt Vonnegut famously wrote, “everything was beautiful and nothing hurt.”

I think back to the “Pope in a puffer jacket” incident in 2023, in which many were duped into believing that the Pope had actually been spotted wearing a designer coat because of some deftly crafted Midjourney images. What made this such a perfect example is that: a) there were minimal actual stakes to this, and b) it played on existing biases about hypocrisies in the Catholic church. Thanks to Sora 2, this kind of blurry nothing event is going to become much more common, the net effect of which is that we’ll waste our time and feel dumb when we inevitably get duped.

But there are actual harmful dimensions to this, too. We’ve already seen how deepfaking is being used in ransom attacks and account access. With Sora 2 and tools like it, all a bad actor needs is a few seconds of somebody’s likeness and a whole host of options opens up to them, from scams that cause family members to pay money because they believe a loved one is in trouble, to gaining entry into a user account using deepfaked content to successfully “trick” two-/multi-factor authentication services.

Sora 2 isn’t the only player here, but what concerns me is the simple fact that Sora 2 is the best publicly available option for these kinds of efforts. At a minimum, I’d encourage you to establish passwords with family and friends now, so that if you or they are the viction of these scams (as they inveitably increase), you have a built-in mechanism by which to determine if they’re real (and remember, statistically speaking, they’re not!).

If you still plan to use Sora 2, I’d ask you to consider a basic ethical economy of use. Does your individual usage contribute to “good” in the world? Do the benefits of your synthetic video outweigh the fact that your use will encourage others to make their own? Are you clearly framing how and why you use these tools to embody your ideals of ethical use?

Atlas and the Dangers of ‘Agentic Browsers’

And then we have Atlas, OpenAI’s new ‘agentic browser,’ which as I mentioned, is even more concerning to me than Sora 2.

To understand how these work, just imagine if you had a helpful AI agent embedded in your browser to undertake tasks on your behalf. With Atlas, ChatGPT is built in, so you can ask it to handle tasks like manual entry (e.g., assemble a spreadsheet using data from this article), research (e.g., find ideal sources of information on a given topic), and composing summaries of articles.

It might sound nice, but under the hood, this is a massive security risk for you.

Read more

Bryan Johnson, the centimillionaire biohacker, entrepreneur, and investor who spends $2 million each year in a well-documented attempt to maintain the body of an 18-year-old, has become the mascot for a time-tested fantasy among the Silicon Valley set (especially among transhumanists and accelerationists): that death, like any other problem, can be optimized away. With his strict diets, blood transfusions, wearable trackers, and, uh, boner tracking, Johnson’s attempt to cheat death is unambiguous—and in case it managed to escape comprehension, he’s made the phrase “don’t die” his literal mantra. Unsurprisingly, it’s also the name of the Netflix documentary that chronicles his journey.

In these efforts, he’s also tried to brand “don’t die” as a rallying cry, framing his own outsized spending and lifestyle asceticism as beneficence, the upshot of which seems to be that his crusade is, at root, intended for public benefit. As flawed as this kind of god-complex justification always is among those with too much power, it’d be easier to believe that he actually believed this if he weren’t simultaneously building out a whole product line called Blueprint (again, he’s already a centimillionaire, worth nearly half a billion dollars), and if he’d proven himself to be committed to the public good in any other capacity, or even basic human decency.

For no reason whatsover, I’m including the conversation I had with Taryn Southern on the Urgent Futures Podcast:

But leaving aside the conclusions we’re left to draw about Johnson as a person, and the distorted ideas of people with too much money who—for some reason—never want to invest it in ways that would genuinely help people who urgently need it right now: “don’t die” is itself reflective of toxic late capitalist ideology. Watching the horror-show of 2025 continue to unfold, while working on my book manuscript for How to Survive the 21st Century, I felt compelled to use the momentum behind the adage to propose an alternative, more life-affirming framing.

“Don’t Die,” But Better This Time

Johnson’s version of “don’t die” is a mirror of the system that produced him—a world that demands endless growth, endless self-optimization, endless extraction. It’s not about life in any collective sense. It’s about domination over the conditions that make life fragile. It’s the capitalist death drive disguised as vitality.

Under these conditions, death is treated as an inefficiency. Aging is a bug to be fixed. The human body is an investment portfolio that must yield returns ad infinitum. Johnson’s meticulous self-monitoring isn’t all that different from the way corporations monitor quarterly growth or the way algorithms are geared to chase engagement at all costs on social media—all seeking the same asymptote: infinity. But unless I missed news to the contrary, we still live on a finite planet.

“Don’t die” is peak growth imperative, tailored to the individual body. Let’s say that we could hypothetically develop the right cocktail of treatments to achieve immortality; how exactly would that work on our finite planet? Taken to its logical extreme, it’s just another vector through which inequality will flourish, with the ultra-wealthy hoarding ever more of the finite resources available on the only planet we call home (don’t even get me started on the misguided attempts to “colonize” Mars).

To understand why inequality is directly correlated to collapse, check out this episode of Urgent Futures with existential risk researcher and Goliath’s Curse author Luke Kemp:

So if you’ll indulge me, let’s refract “don’t die.” What if it weren’t about hoarding vitality, but nurturing and distributing it—not for the individual body, but the collective survival of life on this planet?

Here’s the hard truth: while centimillionaires and billionaires try to cheat death, the planet itself is dying before our eyes, and fast. Forests are burning, oceans are acidifying, species are vanishing at a rate unseen in millions of years (to mention just one recent example, you may have missed the news that two species of Florida coral reefs have been declared “functionally extinct”).

Share

The biosphere that allows any of us to live—the thin membrane of conditions that cradles all life—is in a state of collapse. Our obsession with individual survival—something we’re conditioned to pursue under the alienating conditions of capitalism—has distracted us from the larger system that makes survival possible. The “environment” isn’t just a nice thing that we should be sad to lose (though it’s that too), it literally sustains us. We cannot live without a healthy “environment.”

To put a fine point to this, I’ll draw your attention to a study published this summer, which holds that we are on-track to heat the planet a staggering 3°C above preindustrial levels by 2050—less than 25 years from this moment. There is no way to sugarcoat this: sustained surface temperatures this high are a literal death sentence for most living things on this planet, including you.

In other words, these researchers believe we only have 25 years to make sure that we create the conditions by which we all—and all future generations of our species and others—can survive. While life has managed to spring back after the five previous mass extinctions, it’s taken millions of years each time, and the proverbial players are all different.

Here’s a bit more on that particular subject with Peter Brannen:

In that case, the right kind of “don’t die” is ecological, not egocentric. It’s about the survival of the conditions for life, not just the extension of one life at all costs. It’s about the air we breathe, the soil we touch, the relationships that sustain us; it’s about realizing that our bodies are not separate from the world that holds them—that the line between “me” and “environment” is a porous, constant exchange of breath, water, matter, and meaning.

This other “don’t die” isn’t glamorous. It won’t land you sponsorship deals, venture capital funding, or a glossy WIRED profile. It asks for humility, perseverance, and day-in-day-out struggle instead of transcendence. Under this view, mortality isn’t failure to be streamlined away, but part of the cycle of renewal—compost, decay, rebirth. Sure, there might come a time in the distant future when we have the capability and wisdom to consider what the right conditions for living forever might be, but for now we have to focus on building systems that will outlive us: communities, ecosystems, networks of care.

Bryan Johnson’s version of “don’t die,” when scaled up to the planetary level, is the exact same logic of capitalist economics: infinite growth. The economy must never die, even if everything else does—even if this pursuit ruins its “host.” When rivers dry up, we desalinate the oceans. When crops fail, we genetically modify them and find new tracts of fresh soil to then desertify. When the climate collapses, we build high-tech bunkers. The system’s refusal to die is killing everything and everyone else.

So the correct “don’t die” is not denial of individual death, but the refusal of collective extinction. It’s a politics of persistence—the long, often unsexy work of maintaining the fragile balance that allows complexity, beauty, and diversity to continue on.

Imagine if we spent as much collective energy ensuring coral reefs don’t go extinct as one millionaire spends ensuring his mitochondria don’t age (or comparing his nighttime boners to his son’s)—or if the algorithms optimizing ad revenue were redirected to regenerate the species and lands we’ve devastated.

The irony is that the more Johnson and his peers chase eternal life, the more they affirm the system that will guarantee our extinction. They are avatars of a world that cannot stop consuming. They’re not achieving immortality, but metastasis, a cancerous ideology that flourishes so successfully it drives the body to its inevitable end. To embrace the better “don’t die” requires letting go of the fantasy of control, and with it, the misguided but persistent myth that the economy is somehow separate from ecology, that everything is interrelated.

There’s a different kind of transcendence available, waiting for us—not upward, into silicon, but outward, into relation. This is something that Indigenous peoples around the world have known for millennia: that you become part of something larger not by conquering it, but by belonging to it; that you find “immortality” not in the endless extension of the self, but in the continuity of life itself, even when you’re no longer individually expressed within it.

Johnson calls his brand “Blueprint,” but what he doesn’t seem to realize is his blueprint is one of individual success that destroys the collective. In a sense, I feel sorry for him—this is such a simple thing to grasp, and would open him up to experience genuine meaning and connection in his life. If I had hundreds of millions of dollars, I would joyfully use it to minimize suffering among humans and our fellow earthlings. I don’t mean this in some hokey faux-altruistic sense: it would be the honor and joy of a lifetime to be able to foster change this way. And I suspect you, dear reader, would do the same. The choice not to do this is indicative of a deeply unhappy, lost person who, in the end, will never know peace—even if they manage to live well past 100. We’ll never have his level of wealth, no, but we already know that living more isn’t worth a whole lot if you don’t actually live well.

So yes, absolutely: don’t die. But not as a call for each of us to individually live forever—for this whole weird-ass miraculous pageant to persist. Scientists believe we live in the most biodiverse epoch this planet has ever known, and our economic systems are rapidly unraveling it. That, to me, is a tragedy. In the face of our impending doom, I join so many before me in embracing “don’t die” not as a personal project, but a planetary one.

The last section of this post is only available for paid subscribers. If you’d like to read it and all the other articles in the Reality Studies archive to help you decide if you’d like to become a paid subscriber, I encourage you to use the 7-day free trial option:

Get 7 day free trial

The last two years of U.S. economic outperformance have been unusually concentrated in one sector: artificial intelligence. But AI exuberance isn’t just a stock-market story; it’s a capital expenditure (capex) boom that has yanked growth onward and upward in everything from construction to utilities, semiconductors to cloud computing to venture funding.

To give you a sense of just how much of America’s growth is pegged to AI, here’s a quote from Paul Kedrosky in a July column in the Wall Street Journal:

[O]ne explanation for the U.S. economy’s ongoing strength, despite tariffs, is that spending on IT infrastructure is so big that it’s acting as a sort of private-sector stimulus program.

Capex spending for AI contributed more to growth in the U.S. economy in the past two quarters than all of consumer spending, says Neil Dutta, head of economic research at Renaissance Macro Research, citing data from the Bureau of Economic Analysis.

Understandably, this has caused many analysts to worry about the formation of an AI “bubble,” a term used to describe the phenomenon where the price of a given asset increases far above its actual value, often driven by speculation and investor euphoria. This was the case with the so-called “dot-com bubble,” in which then early Internet companies were overvalued, and ultimately crashed, as well as the 2008 housing bubble, often called the “subprime mortgage crisis,” which resulted in the Great Recession.

As with the aforementioned, if we are now witnessing an AI bubble, and that bubble bursts, the blast radius will not be contained to Silicon Valley. It will ripple through the U.S. economy, creating the kind of cascading slowdown that I believe will feel a whole lot more like a full-on crash than a small “contraction.”

To be clear: I’m not arguing that we ought to keep propping up a prospective AI bubble. I’m attempting to map what I perceive to be a looming risk on the near-ish horizon based on my foresight training, and offer strategy recommendations based on that risk. But if you want a good sense of how I think you should be thinking about AI in society, this is a topic I cover frequently on the Urgent Futures Podcast, including these episodes:

Is it Actually an AI Bubble?

A bubble is when speculation exceeds intrinsic valuation; expectations race ahead of realizations—prices, investment and financial conditions are set by what people think will happen, not necessarily what has happened. Bubbles pop when these bets don’t pay off—when the hype can no longer sustain the lack of real-world evidence, causing investors to scramble for the escape hatch to minimize their losses.

It should be noted that there remains ongoing debate about whether or not the AI hype in fact constitutes a bubble, as described by Rogé Karma in an essay for The Atlantic:

[E]vidence is piling up that AI is failing to deliver in the real world. The tech giants pouring the most money into AI are nowhere close to recouping their investments. Research suggests that the companies trying to incorporate AI have seen virtually no impact on their bottom line. And economists looking for evidence of AI-replaced job displacement have mostly come up empty.

None of that means that AI can’t eventually be every bit as transformative as its biggest boosters claim it will be. But eventually could turn out to be a long time. This raises the possibility that we’re currently experiencing an AI bubble, in which investor excitement has gotten too far ahead of the technology’s near-term productivity benefits. If that bubble bursts, it could put the dot-com crash to shame—and the tech giants and their Silicon Valley backers won’t be the only ones who suffer.

For my part, I do believe this is a bubble—with all requisite disclosures that I’m not an economist and nothing I write here is financial advice—the primary questions I have at this point are: how big has that bubble already gotten? How much bigger will it get before it crashes? What will a crash mean for the U.S. (and therefore global) economy—especially given the economic turbulence that has already been created by the Trump administration?

Investors seem to be pricing in a future of unbroken demand for chips, data centers, and generative AI applications, and valuations have reached astronomical levels. Nvidia trades at ~54 times its actual earnings, while Microsoft, Alphabet, Amazon, and Meta are investing at capex levels that dwarf historic norms. Venture capital has followed suit, with AI absorbing 57.9% of all global venture capital funding in early 2025 (read that stat again) and a growing share of venture debt. The stock market itself has become hostage to this concentration: the “Magnificent Seven” dominate the S&P 500, with Nvidia alone accounting for close to 8% of its weight. Most alarmingly, a recent Financial Times report—ominously titled, “America is now one big bet on AI,” found that the hundreds of billions of dollars companies are investing in AI account for an eye-watering 40% share of all U.S. GDP growth in 2025.

This is precisely the setup that has defined past bubbles—prices, flows, and entire indices tilting on the assumption that a single technological narrative will keep delivering.

Again, it’s important to remember that this isn’t just about markets in the abstract. AI has become the engine of U.S. growth in measurable ways. National accounts data show that business investment in IT equipment, software, and data-center structures accounts for almost the entirety of U.S. GDP growth in the first half of 2025. Data-center construction has surged to annualized highs of around $40 billion, a figure unmatched in U.S. history, while power consumption forecasts from both the EIA and IEA point to AI-driven loads reshaping America’s electricity mix. Hyperscalers, large-scale cloud service providers that operate massive data centers and networking infrastructure, are spending tens of billions each quarter to meet demand, with combined capex from Microsoft, Alphabet, Amazon, and Meta projected to reach $350 billion this year and $400 billion in 2026.

Even household consumption is being buoyed by AI. The stock market wealth effect is no longer broad-based but heavily tethered to the outperformance of AI-linked firms, which have accounted for the bulk of index-level returns. In short, the U.S. growth “beat” relative to other advanced economies has coincided directly with the AI investment supercycle.

Thus, a reversal would be nothing short of cataclysmic in the cross-category cascade it would likely induce. If hyperscalers or enterprises slow AI investment, the contraction would strike several sectors at once: construction, semiconductors, utilities, electrical equipment, and software. The stock market would feel the effects almost immediately. With so much wealth concentrated in a handful of companies, a correction in AI-linked names would drag retirement accounts and passive investment vehicles down with it, eroding household net worth and weakening consumption. At the same time, the financial plumbing behind the boom is fragile. Much of the AI buildout has been funded through vendor financing, leasing arrangements, and debt collateralized by future compute contracts. Should resale values of chips fall or demand forecasts slip, weaker firms in this ecosystem—especially niche data-center builders and cloud intermediaries—would face credit stress that could spread more broadly.

The energy sector would be caught in the whiplash as well. Utilities and grid operators are rapidly expanding to accommodate expected AI demand, with some forecasts (including by the U.S. Department of Energy) projecting data centers consuming roughly 10% of U.S. electricity within a few years. If that demand under-delivers, capital-intensive grid and generation projects may no longer pencil out, leaving stranded assets and stalling the broader “reindustrialization” narrative that policymakers have been celebrating. Venture capital and startups would also feel the shock. With the aforementioned 57.9% of global VC dollars in early 2025 flowing into AI and venture debt following the same path, a re-rating of AI expectations would leave other sectors starved of capital, depressing growth and employment in startup-heavy regions.

Share

How the Bubble Could Burst

The potential triggers for such a downturn are not hypothetical. If enterprises fail to realize near-term productivity or revenue gains from AI adoption, investment will slow. If the global supply of GPUs normalizes, undermining the economics of compute resale and leasing models, financing structures will unravel. If grid bottlenecks or regulatory interventions delay deployment, the returns on massive data-center spending will fall short of expectations. Any of these could be enough to pop the bubble, and together they make the current trajectory unusually vulnerable.

In an August post on The Rip Current, explains why he sees early indicators for these types of triggers. He argues that OpenAI’s recent insistence on the real-world value in customers’ lives and practical application(s) during the launch of their flagship large language model, GPT-5, signals increasing pressure on the AI giant to start delivering actual return-on-investment after a few years of loss-leading hype:

The Rip Current by Jacob Ward

What GPT-5 Really Represents

OpenAI described its newest model as a pocket full of PhDs, a cancer doula, and a eulogy writer. But behind the headlines are the enormous pressures building on this industry. Here’s a breakdown of what we learned today…

Listen now

9 months ago · 26 likes · Jacob Ward

In a recent episode of the Urgent Futures Podcast we recorded (which will be released in a few weeks!), he further expressed alarm about the recent announcement that Nvidia would be investing $100 billion in OpenAI—this type of circular seller-investing-in-customer entanglement only amplifies risk.

The chain reaction is easy to picture. Earnings disappointments from the AI leaders would translate into weaker guidance on capex, which in turn would hit semiconductor makers, cloud providers, and infrastructure firms. Stock prices would follow, dragging household portfolios lower and dampening consumer spending. Canceled construction projects and delayed utility expansions would ripple through employment and industrial output. Credit conditions would tighten, first for AI startups and then for the broader venture ecosystem. What begins as a technology correction could then cascade into a broader economic downturn.

Skeptics argue that this cycle is different, pointing to genuine earnings strength at firms like Nvidia and to credible long-term productivity potential. That much could be true: generative AI may well transform industries over a decade or more (in ways good, bad, and ugly). But markets, utilities, and builders are not priced for gradual gains. They are priced for explosive, immediate, and sustained demand. If the payoff curve proves flatter in the near term—as history suggests it often does—the economic impulse turns negative just as quickly as it turned positive (if not quicker).

What to Watch

The signals to watch are straightforward. Capital expenditures guidance from hyperscalers for 2026 and beyond will reveal whether they see momentum continuing. Data-center construction figures, which have so far only risen, will show whether the building boom is peaking. Power-demand revisions from official agencies will highlight whether the AI load is truly materializing at scale. And stock-market breadth will continue to serve as a warning: if the health of the U.S. economy rests on the performance of two or three companies, the foundation is almost comically thin.

We must remember that America’s recent economic resilience has not been the product of broad dynamism but of a singular bet on AI. That bet has supported GDP, construction, energy, finance, and household wealth—but it has also left the U.S. economy balanced on a knife’s edge. If the bubble pops, it will bring the entire American growth story crashing down with it—at an already perilous moment in geopolitics, thanks to President Trump’s disastrous economic and trade “policies.”

What Can We Do to Collectively Prepare?

If the past 2-3 years have been a high-wire act balanced on the AI industries hype and promises, the next phase will determine whether the U.S. economy builds a safety net—or keeps walking along the tightrope without one. There are no simple fixes for a bubble this large and intertwined with real economic activity, but there are several steps policymakers, investors, and firms can take to soften the landing if expectations deflate.

As with so many things in our lives, diversity is the answer. In the case of the prospective AI bubble, investors ought to diversify sources of economic growth. (We’ll leave aside my own gripes about America’s obsession with endless growth for a moment). The concentration of capital, talent, and policy attention around AI has starved other sectors of investment. Redirecting some fiscal incentives and credit flows toward neglected areas—here I’ll bang my drum for more robust investment in housing, public infrastructure, and clean energy—can reduce dependence on a single technology. Economic resilience depends on breadth, not just depth, such that future damages in any one sector remain (more) localized.

We also have to weaken the feedback loop between financial markets and infrastructure build-out. Current utility planning, grid expansion, and chip-fabrication decisions are being made under the assumption of uninterrupted AI demand growth. Regulators and utilities should adopt scenario planning that includes lower-than-expected demand trajectories, to avoid stranded assets if forecasts don’t materialize. The added benefit here, of course, would also be less of a drain on energy infrastructure and less harm to the environment in the form of greenhouse gas emissions and drinking water waste (for cooling data centers).

As a foresight strategist, I’m strongly of the mind that we must improve the quality of demand forecasts and public data. The speed and opacity of AI-related build-out have outpaced the ability of agencies like EIA and FERC to provide timely data. A more transparent and standardized reporting framework for data center electricity demand, construction pipelines, compute investments, and greenhouse gas emissions would help both the public and private sectors make more realistic plans.

Obviously, we also have to build countercyclical buffers now, which will constitute a far less costly use of our time and money than “clean-up” after the bubble pops. Corporate treasurers, venture capitalists, and policymakers should treat today’s boom as temporary windfall revenue, not permanent baseline. That means resisting the temptation to ratchet up permanent spending commitments and instead using some of the gains to shore up balance sheets, invest in productivity-enhancing public goods, or pay down debt. Fiscal space and financial stability will matter enormously if the tide turns.

Most broadly, we all must recognize that the narrative itself shapes outcomes. We have to get clear-eyed about what AI actually can and cannot do now and in the immediate future, rather than obsessing over far-future hypotheticals. Much of the AI boom has been driven by expectations of explosive returns—on the back of promises made by companies and investors who stand to directly benefit from said boom. We have to temper that narrative to reduce systemic fragility. It’s already clear that, while AI may hold promise as a transformative technology some day, it is not currently living up to demand, creating far fewer real-world examples of genuine innovation than the tech oligarchs have been proclaiming are right around the corner for the past three years. If we get really honest with ourselves, we’ll be better equipped to act thoughtfully. There is no reason that AI needs to be developed at breakneck speed—and moreover, a whole lot of reasons why it shouldn’t be.

As much of an advocate for degrowth and post-growth as I am, I have to point out that if we retreat in panic as rapidly as we surged in hype, we’ll pop the bubble, which will indeed yield degrowth but in a fashion that harms billions of people. So as much as AI critics may want us to abandon the technology altogether, that’s not actually realistic under the lens of systemic wellbeing—at least in the near term.

Instead, what we need is a prudent ramp down in speed, to build a more sensible foundation upon which to develop the technology using the infrastructure (of various kinds) that has already been built. If the bubble does burst, as bubbles tend to do, this kind of preparation could mean the difference between a tummy ache and a full-blown global economic crisis.

What Individuals Can Do to Prepare

Bubbles are collective phenomena, but their aftermath is experienced individually. If the AI boom pops, the effects will ripple through household wealth, job markets, and local economies. It’s impossible to fully insulate yourself from such a macroeconomic shock, but there are practical steps you can take to build resilience now.

Read more

On March 2, President Donald Trump announced by “truth” that a crypto reserve which would feature the cryptocurrencies Ripple (XRP), Solana (SOL), and Cardano (ADA). In a later post he added that Bitcoin (BTC) and Ethereum (ETH) would be included as well. This was historic news—I knew it immediately, and said so in my flash analysis that day.

As more has been revealed—including the White House Crypto Summit on March 7—I’ve had a chance to conduct a deeper analysis. What follows is designed for those with no understanding of crypto to grasp the basic ideas. To clarify: I’m not going to explain how blockchain technology works other than where necessary, but if you want an accessible explainer, I like this one from Reuters.

Some background for folks who may not know: I’ve been covering emerging technologies, including cryptocurrencies, since 2015. In Decemeber 2017, my articles about cryptocurrencies in Forbes accounted for 5% of all of Forbes.com site traffic. I curated the first-ever museum retrospective on the history of NFTs, and most recently worked as Arts & Culture advisor for Protocol Labs, an open-source research & development company whose key projects include IPFS and Filecoin.

Okay, let’s get to it.

What is the Trump Crypto Reserve, and What Has Happened so Far?

In his March 2 announcement, President Donald Trump framed the crypto reserve as a move to “elevate this critical industry after years of corrupt attacks by the Biden Administration” and to make the U.S. “the Crypto Capital of the World.”

• Though it hasn’t been implemented yet, other than the Bitcoin reserve described below, such a reserve would in theory use taxpayer dollars to buy the digital assets named above, marking a dramatic shift in U.S. financial policy (to put it lightly).

• Critics immediately noted that this initiative “poses countless conflicts of interest—especially for Trump himself,” essentially amounting to a taxpayer-funded gift to the crypto industry.

• This move appeared to make good on then-candidate Trump’s July 2024 campaign promise to create a “strategic national bitcoin reserve.”

The announcement triggered a trading frenzy, driving up the named coins’ prices (Bitcoin jumped over 10% and Cardano surged by nearly 70% that day), though they have all since fallen below their early March prices.

Executive Order

On Thursday, March 6, President Trump issued an executive order establishing a government stockpile of bitcoin. The move is more for show than anything; under the EO, the U.S. government will simply retain the estimated 200,000 bitcoin it has already seized in criminal and civil proceedings, according to Trump’s “crypto czar” David Sacks (who we’ll be discussing in more detail below).

• With this EO, the White House appeared to adjust course, holding the bitcoin-only strategic reserve alongside a separate crypto stockpile comprised of ADA, ETH, SOL, and XRP.

• In a March 7 episode of the All-In podcast, Sacks explained that the Bitcoin reserve would be moved “to the digital stockpile,” and that “the purpose of the stockpile is responsible stewardship, it’s a place for safekeeping, it’s a centralized account under the direction of the secretary of the Treasury and the secretary of the Treasury will figure out how to maximize the value of these holdings.”

White House Crypto Summit

On March 7, the White House hosted the first-ever “crypto summit,” bringing together key industry figures from Coinbase, Ripple, Gemini, Kraken, Chainlink, and Robinhood, emphasizing the administration’s intent to foster growth and innovation in the sector through clearer regulatory frameworks.

• “Last year I promised to make America the Bitcoin superpower of the world and the crypto capital of the planet, and we're taking historic action to deliver on that promise,” Trump said in his address.

• Trump also confirmed his commitment to bitcoin—seeming to elevate it to a position above other cryptocurrencies—in a reserve described as a “digital Fort Knox,” in a nod to the U.S. gold reserve.

Advocates, epitomized by the industry leaders in attendance at the summit, believe this crypto reserve is a chance to fortify U.S. economic supremacy worldwide.

• “My job is not to encourage people to buy crypto. My job is to create an innovation framework for the United States,” Sacks said, adding that “we’re only allowed to buy more [cryptocurrencies] if it doesn’t add to the deficit or the debt or cost taxpayers,” seeming to address the concerns that taxpayer dollars would be used to build either the Bitcoin reserve or crypto stockpile.

Trump also later framed the move as a means of competing with China in tech.

• “China is pushing forward very strongly as usual, but we're way in the lead, as we are in AI and other things, and we want to stay there,” Trump said.

Crypto Contraction

The initial trading frenzy after the March 2 announcement has died down—and the crypto market has continued to tumble, falling to its lowest market capitalization since the post-election surge.

Agne Linge, head of growth at crypto platform WeFi, told Forbes that the more constrained approach of the bitcoin reserve may have cooled short-term enthusiasm, while leaving the door open for long-term growth through geopolitical implications:

“The subpar strategic bitcoin reserve move triggered the uncertainty in the crypto market…. With the executive order directing agencies to consolidate seized bitcoin to form the reserve, crypto investors see the move as a trick, as no new bitcoin purchase was announced. Despite the knee-jerk reaction from investors, the fact is that the bitcoin reserve mandate authorizes the acquisition of bitcoin through means that will not cost taxpayers’ money. The available options in this regard include bitcoin bonds and the sales of its gold reserve to fund more purchases. In the long term, the bitcoin reserve shift might benefit the coin. This thesis hinges on a possible race it has triggered that may see other sovereign nations make similar moves.”

Blockchain Technology: Ideals vs. Realities

Blockchain and cryptocurrencies are not de facto bad or evil. The baseline of cryptography, distributed ledgers, and decentralized consensus protocols is rooted in worthy ideals—namely that of putting power in the hands of collectives rather than the centralized individuals and entities (which increasingly concentrate power).

Unfortunately, the confusing and unregulated “wild west” aspects of the technology also make it an ideal zone for graft, fraud, and ransoms (e.g., cyberattacks on hospitals frequently involve ransom payments in cryptocurrencies because they are easier to send without oversight and harder to track and retrieve once disbursed).

I resonate with something that Urgent Futures podcast quipped, which is that the more boring the use case, the more likely blockchain is good for it.

As authoritarian regimes intensify around the world, for example, blockchain technologies could prove invaluable for preserving evidence of war crimes for posterity—a use case demonstrated by Starling Lab.

But there’s two sides to that coin. Cryptocurrencies also offer sneaky and expedient means for committing crimes for those very same authoritarian regimes.

Conflicts of Interest and Self-Dealing

Following the March 2 announcement, observers quickly pointed out that Trump’s crypto reserve could enrich Trump and his family.

Paid subscribers can access the following content below:

• Conflicts of Interest and Self-Dealing

• Why is the Crypto Reserve Ripe for Corruption?

• Hypothetical Abuse Scenarios of the Crypto Reserve

• Systemic Issues: Political Corruption with Crypto

• Who is David Sacks? Trump’s “Artificial Intelligence and Crypto Czar”

• How Trump Allies and the Crypto Community Responded to the Trump Crypto Reserve

• Comparisons to Past Precedents

Read more

It’s mayhem out there, y’all. Trump just announced a strategic crypto reserve.

Even if you hate talking about crypto, it’s important to understand the ramifications of this announcement. I’m sketching my initial thoughts below, which I’ll be updating as I learn more. But suffice it to say this is a huge deal. Some are calling it the most corrupt act ever taken by a sitting U.S. President.

This post is no longer free. Subscribe now so you always get free posts straight to your inbox while they’re available.

Subscribe now

Paid subscribers make analysis like this—ranging from AI and ClimateTech to polycrisis and post-growth economics—possible. I hope you’ll consider becoming a paid subscriber today.

Read more

Ted Chiang is one of the most important writers of our time, not only for his fiction but for his essays, recently on the subject of machine learning/artificial intelligence. “ChatGPT Is a Blurry JPEG of the Web” and “Will A.I. Become the New McKinsey?” in the New Yorker are two of the most lucid public-facing essays about the intersection of AI, creativity, and culture. I can’t recommend them enough.

Which is why Chiang’s latest essay, “Why A.I. Isn’t Going to Make Art” feels so bewildering. As with his previous essays, it deftly identifies how late capitalism perversely incentivizes AI’s potential shortcomings, amplifies harmful progress narratives, and stands to damage art and art’s role in society—but in the process it manages to misrepresent both AI and art, seeking to freeze frame and mechanize what are both in fact complex, liquid forms. I believe Chiang and I agree on the broad strokes, but disagree on what it means, what it asks of us, and ideal strategy—and what follows is offered in the spirit of dialogos.

[NOTE: this piece doesn’t recap Chiang’s piece, nor does it address every single point he makes. His piece is very worth reading; I encourage you to check it out before coming back here, though it’s not necessary to understand my arguments]

Art is Contextual

Chiang introduces his argument by defining art—which he admits is a notoriously tricky task—as “something that results from making a lot of choices.” Certainly great art involves a lot of choice, but this definition is at best a subpoint for a more comprehensive, explanatory definition. What field does not involve a lot of choices at the level of practitioners and experts? Science certainly involves a lot of choices, as do math, basketball, and cooking. Even design, often held up as the yin to art’s yang, involves a lot of choices.

An example of a definition for art I would offer is: creative expression that attempts to communicate something true, honest, and/or beautiful through specific materials, in such a way that it could not be communicated otherwise. To be clear, this can only account for one idealized version of art based in my own subjectivity, and, as will be discussed below, the human realities of what we collectively call art frustrate any attempt at a single definition.

We might overlook Chiang’s misrepresentation if it weren’t the foundation upon which he develops his argument. Outside of definitions and semantics, art is not a fixed thing. It evolves alongside humanity’s symbol- and tool-making capabilities.

Painting on canvas, for example, only came into existence in the 14^th century, and the stretched canvas (the canvas as we understand it today) was only invented in the late 19^th century. We think of poetry today as a written medium, but some scholars believe it was an oral medium first, predating literacy. What was expressed as poetry then is anathema to much of what is produced now, and even the modes of oral poetry that persist today are vastly different in style and function than the oral poems from thousands of years ago.

Thus, saying AI will never create art implicitly impoverishes futures of what art might be, modalities of artistic expression that do not exist today. Even carrying the assumption that generative AI will remain static and only ever produce outputs that are blander than human-produced art, there could be future milieus in which producing the blandest, most generic output is regarded as the highest expression of art. In this case we might not say that the AI model is an artist, but if, as today, art is deemed as such by a loose consensus of key stakeholders (artists, institutions, distributors, financiers, enthusiasts, public audiences, et al.), it’s art.

Subscribe now

Artists Have Already Made Art with AI (and Have Been Doing so for Decades)

Though the stated thesis of Chiang’s argument addresses question of a model being able to autonomously produce art, his evidence relies on examples of humans using generative AI in the production of art, a critical difference. In this, he exposes himself to easy rebuttals; of course artists can make art with AI, and indeed have been since the 1960s. Rather than burden this essay with the abundance of potent examples, I’ll instead point to the exhibitions I’ve curated featuring artists doing exactly this.

A wonderful dialogue with artist and science communicator , which examines AI in connection with culture, power, mathematics, and dimensions…also the ways that coral reefs exhibit the curved structure of spacetime and more:

There’s an inevitable counterargument here that what constitutes capital-A art is rooted in subjectivity, and perhaps the artworks selected above won’t read as art to certain audiences. Capital-A art as we understand it today lives within the lineage of Modernism, epitomized by Duchamp’s readymades, ordinary objects that Duchamp “elevated” as art by the simple function of his artistic choice. However true this reality may have been before, readymades rendered it mainstream, diffusing this relationship to art and materials so thoroughly into the establishment that it’s now hard to imagine how one would even produce, discuss, and critique art absent this commitment to concept. If materials as seemingly far-fetched as a bike wheel, urinal, thread, and a wheat field can be used to create some of the most iconic works in modern art history, it almost feels silly to state that AI could as well—even including the most boring, generic large model, to say nothing of the handcrafted, bespoke models that artist-technologists might create.

Chiang’s argument that generative AI incentivizes lazy participation is legitimate, and here I want to echo his claim that AI will probably have a large-scale blandifying effect on creative communication. This is especially true as it pertains to the periods of early learning in one’s artistic practice, which are the hardest and often most thankless because aspiring artists’ tastes so outpace their capability. It is here that critical learning occurs, and it’s not hard to imagine the easy allure of generative AI undercutting this necessary, formative learning.

In an earlier essay, Chiang nails this:

“Sometimes it’s only in the process of writing that you discover your original ideas. Some might say that the output of large language models doesn’t look all that different from a human writer’s first draft, but, again, I think this is a superficial resemblance. Your first draft isn’t an unoriginal idea expressed clearly; it’s an original idea expressed poorly, and it is accompanied by your amorphous dissatisfaction, your awareness of the distance between what it says and what you want it to say. That’s what directs you during rewriting, and that’s one of the things lacking when you start with text generated by an A.I.”

But arguing that AI will interfere with respective individuals’ artistic development is a far cry from claiming AI will never (be used to) create art. Thus, my claim is not that the diffusion of large AI models into mass culture will be a net good for art and society, but rather that future societies are fundamentally unknowable, and we betray a limiting bias when we attempt to project today’s value systems (artistic or otherwise) onto them.

Large Commercial Models Are Only One Kind of AI—and Only a Few Years Old

Generative AI as we understand it today might be a poor tool for producing art in most cases, and have knock-on deleterious effects on artist development, but claiming it will never (be used to) create art discounts what they might evolve into, as well as the prospect that AI models themselves might be art.

Chiang seems to want to make claims about AI as a monolith while limiting the scope of his argument to contemporary large commercial models. Given the outsized presence of AI in public discourse, it’s easy to forget that generative AI really only became a public discussion in 2022 through the seemingly sudden capabilities of Midjourney, DALL-E, and ultimately ChatGPT. There had been buzzy moments prior, such as the spate of “an AI wrote this article wow” pieces produced using GPT-3 in 2020, but 2022 was when “generative AI” became the category we now understand it to be.

As such, the claim that there exists no artistic potential feels near-sighted and epistemically brash, betraying a disposition about how it feels for AI to creep into everything rather than a more measured assessment of what it might mean for the futures of art, creativity, and culture. The transformer deep learning model that underpins ChatGPT and many other LLMs, was originally proposed as a translator, and its capability came as a major surprise to its inventors. Even taking the longest view to its initial proposal in 2017, we’re only in the very early phases of what it means that these particular deep learning models are the ones that exploded into the public. There are countless other existing and forthcoming types of models based on different frameworks/proposals, such as the state space model (SSM), which could be years away from being well-understood and/or able to radically alter AI’s capabilities. Thus, if the option is between predicting that AI will never create art and that some future version of AI will create art, statistics would side with the latter.

And the culture of AI is vaster than just the large commercial models. The hype of the past couple years has also set in motion breakneck startup investments, social and entrepreneurial dynamics, and academic research projects, and critically, galvanized the open source community. These too are lagging indicators—they take time to gain recognition.

In some cases it won’t be the individual outputs that will be regarded as the art, but the models themselves. This notion extends arguments that digital artists have been persuasively making regarding generative art (not to be confused with generative AI), in which they claim that it is the code that is the true art, and the outputs (i.e., individual images) are how that art is expressed.

What happens when AI capabilities (however successful they are or are not) intersect with neurotechnology, especially brain-computer interfaces?

Dipping into more fantastical speculation: imagine if the initiatives to use AI to translate animal speech succeed. What new hybrid forms of art and communication might emerge in such a moment? More importantly, what are the new dangers and possible sources of harm that this development will bring—and who do we imagine will be the ones to identify and communicate those to the public?

Why Does This Matter?

Perhaps the AI of our futures will lean more toward perfunctory tasks than artistic ones, but it’s important to acknowledge the role artists play as translators of their respective paradigms and zeitgeists. This artistic translation might not merely be for the general betterment of society, but identify key points of intervention in problematic structures, which the public may not yet perceive or understand. As Elan Mastai said, “When you invent the car, you also invent the car accident. When you invent the plane, you also invent the plane crash.”

What are the proverbial accidents and crashes latent in AI and the surrounding culture that artists might identify, communicate, and possibly intervene in?

Modern digital life involves participating with technologies that have been engineered under the incentives of late capitalism (or technofeudalism, depending on your preferences). They involve high degrees of surveillance, targeting, and psychological manipulation, and for many, they form the bedrock of daily interactions. Art is only one possible means of intervening these systems, but it’s an important one. Artists who intimately understand the material of AI will be more capable of levying critique and obstructing the project of extraction.

This is something I discuss in depth with & Caroline Sinders in a recent episode of Urgent Futures, and I can’t recommend their wisdom enough:

Thus, this isn’t an esoteric industry debate. The narratives underpinning the development of AI are potent, driving industry investment, the future of labor, and energy realities. It remains popular, especially among artists, to reject AI outright, but this entails an unfortunate resistance to developing literacy and awareness about what is going on among the very people who are possibly the most poised to effectively intervene. I fear that pieces like “Why A.I. Isn’t Going to Make Art” confirm that stance, as if to say, "Nothing to see here, folks!" Even though that disposition may prove appropriate in many cases, especially regarding the production of subpar, concept-less images that parade as art, this attitude downplays the role AI will continue to have in society, and the need for liberatory positions and prosocial interventions in the face of its ongoing, breakneck development.

It feels weird doing a piece like this, as Ted Chiang is something of a literary hero for me. His stories and essays are superlative examples of the role artists can play in shaping discourse about emerging technologies. I offer this essay in the tradition of his earlier essays, as I believe these are important considerations not just within art and AI, but for society, power, and policy. If he happens to read this and wants discuss it all on Urgent Futures, I’d welcome the opportunity. I continue to have immense respect for his overall clarity of vision in how AI will be used to flatten discourse and serve the interests of the wealthy at the expense of the many.

Yes, I see opportunities for artistic invention with AI, but I see many more opportunities for AI to be used to exacerbate the worst of modern society. In this I’m in total agreement with Chiang, but in this case we disagree on how this should be expressed. Given how vast and nuanced the possibility space of AI is, it feels critical to advocate for artistic AI literacy—among both artists and the broader public—and it seems to me that trying to prove how current-generation AI is creatively deficient sabotages that goal.

By now you’re probably aware of the Drake-Kendrick Lamar beef, which has kicked into high gear over the past month. If so, skip the bullet points—but since it’s salient background for the discussion that follows, I’ll quickly recap the main beats:

• Drake and Kendrick have maintained a simmering dislike for each other for at least a decade. Other than subliminal jabs it’s never spilled over into public confrontation.

• Drake releases For All the Dogs in October 2023, which includes lead single “First Person Shooter” with J. Cole. In the song, J. Cole raps: “Love when they argue the hardest MC / Is it K-Dot? Is it Aubrey? Or me? / We the big three like we started a league, but right now, I feel like Muhammad Ali.” The “big three” here refers to the mythology of Drake, J. Cole, and Kendrick being the best rappers of the 2010s. In the same song Drake pronounces himself the GOAT.

• On March 22, 2024, rapper Future and producer Metro Boomin release a collaborative album titled We Don’t Trust You. The track “Like That” features a verse by Kendrick Lamar, who takes direct aim at Drake (and J. Cole…but mainly Drake), saying “Motherf*** the big three, n****, it's just big me” among a host of other direct and indirect provocations.

• Drake, Future, and Metro Boomin have all worked together in the past, but rumors had been mounting since late 2023 that Metro Boomin and Future were beefing with Drake, stemming from perceived disses directed at the producer in For All the Dogs.

• On April 5, J. Cole releases a response track that is ambivalent (at best), prompting this gem from critic Alphonse Pierre’s Pitchfork review: “Who the f*** wants to hear this extremely measured and level-headed assessment of Kendrick’s albums in a diss track?”

• Future and Metro Boomin release another collaborative album on April 12, which includes more words for Drake, this time not from Kendrick Lamar but directly from Future, and broadening to include other former Drake collaborators A$AP Rocky and The Weeknd.

• On April 13, a demo version of the song “Push Ups” leaks. It’s attributed to Drake, but since it didn’t come from any of his official channels, there’s speculation that it might be AI-generated. It’s very much a diss directed at all of the above, as well as Rick Ross, who had unfollowed Drake on Instagram.

• On April 15, Rick Ross enters the fray with “Champagne Moments,” a Drake diss accusing Drake of getting cosmetic surgeries like a nosejob, and includes the much-discussed line, “I know you got your Dockers on with no underwear, white boy!”

• On April 19, Drake officially releases “Push Ups” in its final form, which features stronger production and a few notable lyrical tweaks. On the same day, Drake releases another diss via Twitter and Instagram, titled “Taylor Made Freestyle.” This track will be the focus of the discussion from here.

Note: I encourage you to listen to the various disses. They’re all worthwhile listens, with the exception of “7 Minute Drill.”

Reality Studies Recommends:

NOTE: Purchasing through these links supports the work I do with Reality Studies. I will only ever share products that I would endorse regardless of financial incentive.

ZBiotics

Right now, get 10% off ZBiotics. Just head over to zbiotics.com and use code JESSEDAMIANI. The next day after drinking feels way better when you take one of these. Genetically engineered by a team of PhD microbiologists, ZBiotics is a probiotic drink that breaks down the byproduct of alcohol responsible for rough mornings after drinking (acetaldehyde).

NordVPN

Right now, get up to 66% off + 3 months extra through this link. Some people tell me that “VPN” brings to mind ideas of hackers and the dark web, but honestly VPNs are just an extremely easy way to stay much safer online. I’ve used NordVPN for the past four years, and appreciate what they offer, including Threat Protection against malware, 24/7 customer support, fast speeds, and more. One account can protect up to 6 devices (phone and computer), and they don’t track or share what you do online. Another benefit: you can always access the content/apps you have at home, wherever in the world you are.

“Taylor Made Freestyle” and the New Iteration of Drake AI Weirdness

Last year, a couple AI Drake covers—namely “heart on my sleeve,” by an TikTok user ghostwriter977—went viral. Listen here if you haven’t heard it. Note: what “AI cover” generally refers to is someone using AI voice tools (and in some cases AI instrumentation) to produce work “as” a given artist. In other words, a human writes the song and sings/raps, using a voice clone to mimic the style of the artist being “covered.” Drake is one of the most well-known pop stars in the world; naturally his work would be an early target for such creative experimentation.

“Heart on my sleeve” sparked conversations about the future of IP and creative ownership in a context of ubiquitous generative AI tools. There have been plenty of fun AI covers since then—a personal favorite is “Somebody That I Used to Know” in the style of Linkin Park—but, as tends to happen, as the tools have become more accessible and general knowhow increases, novelty fades. By and large, these now operate more as memes than meaningful competition for the respective artists’ bottom lines. Artists and platforms are figuring out how to label synthetic content and how to foster community participation through collaborative economic models.

“Taylor Made Freestyle,” though, is by a major artist. Drake does appear in the song, but not until we’ve heard from AI versions of 2pac and Snoop Dogg. The two icons of West Coast hip hop exhort heir apparent Kendrick Lamar to rise to the occasion and respond to Drake’s “Push Ups.” For his part, Snoop Dogg (the human being) handled the whole thing with the Internet equivalent of an eyeroll, saying, “They did what? When? How? Are you sure? … I’m going back to bed,” in an Instagram video.

Check out the latest episode of the Urgent Futures Podcast, featuring Eric Czuleger, author of You Are Not Here: Travels Through Countries that Don't Exist:

Given that, Drake may not find himself in legal peril (though, hey, still plenty of time!), but this form of AI recreation sparks copyright, fair use, and other ethical considerations. No, Drake isn’t monetizing the song, but it’s hard to argue that using this tactic won’t yield any indirect economic benefits—streams, increases in followers on social media, press coverage, interest in investment properties, et al. Is it parody? I guess you could argue it is—but then you could also meaningfully argue that it’s not. What data was used to train the voice models? Did Drake use an existing voice model of these artists, or did he/his camp train and finetune them?

I also find myself wondering about the creative implications. Drake used the fact that he’s the subject of AI covers, and the fact that many wondered if his own work were in fact an AI cover, and flipped the script. Puckish as the prank may be, there is real creative innovation here. The silly and almost throwaway aspect of 2pac and Snoop Dogg’s features in the track likely insulate Drake from legal blowback, and simultaneously hack the game of discoverability on social media. This is designed (ahem, tailor-made, if you will) for a public spat of this nature, but there are plenty of other use cases in which we might imagine guest features—ones that are endorsed by the respective artists or by the estate/rights holders (if the artist is deceased, as in the case of Tupac Shakur). And let’s not even go too far down the road of what this might mean for blockbuster biopics and jukebox musicals…

We were already inevitably going to see synthetic songs and guest features by artists of yore, but I suspect we’ll be seeing a lot more a lot sooner thanks to this diss track. I wrote about how AI interacts with genre last year, and I’m wondering about a lot of the same things right now—how exactly will this zombiism impact audience expectations in art and creativity (in music and beyond)? What aspects of artists’ outputs will become the “signature” elements that get reproduced, and how will the reproductions impact them as artists and the broader development of art forms, aesthetics, and pop culture? Last year’s Wes Anderson AI trend was the jump-off point to the article I wrote on AI and genre. It was a playful edge case that provoked quite a few questions about how synthetic creativity can, might, and will become commonplace. This Drake diss moment feels to me like another such instance.

Subscribe to the Urgent Futures podcast on YouTube:

My latest exhibition, SMALL V01CE, opens at Honor Fraser Gallery this weekend (for those of you in Los Angeles, come say hi! January 13 from 6-8pm.)

Human evolution is often presented as a story of expanding intelligence. In SMALL V01CE I wanted to trouble these assumptions by examining the intersection of artificial intelligence (AI) and instinct, intuition, and feelings. It grew out of a set of questions I’ve been pursuing as part of my research into Postreality, and a desire to understand how artists were thinking through the aforementioned. And wow did they rise to the occasion! It’s been a joy to be in dialogue with them and to see what work they developed or selected to exhibit. The curatorial frame for the show can be found below.

“[C]ultural activity began and remains deeply embedded in feeling. The favorable and unfavorable interplay of feeling and reason must be acknowledged if we are to understand the conflicts and contradictions of the human condition.”

—Antonio Damasio, The Strange Order of Things

The hairs on the back of your neck stand up. The basketball player hits the buzzer-beating fadeaway in pure flow. The sudden sensation that you’re being watched. The artist looks at the work and a small voice inside them tells them it’s done—without understanding why.

Human evolution is often presented as a story of expanding intelligence. Indeed, our faculties for learning, recognizing patterns, symbolic thought, and coordinating these insights have been integral factors in what we have become. It follows in this line of thinking that humans organize our lives according to logic and rationality, but studies reveal that instinct, intuition, and feelings are underlying drivers of our choices and experience of the world. In fact, according to some neuroscientists, these biological algorithms predate intelligence, a genetic heritage linking us back billions of years to our bacterial ancestors. In other words, our understanding of the very nature of intelligence is likely based on incomplete ideas and flawed assumptions.

Now, a new form of intelligence is capturing the public imagination: artificial intelligence. In specific, a category called “generative AI,” which refers to a number of deep learning techniques capable of producing outputs like images, videos, and audio—forms we often associate with art and creativity. Machine intelligence is already weaving into creative making and tooling, a trend that appears to be accelerating. The explosion in capability among adjacent technologies like autonomous vehicles (drones), processors (GPUs), and sensing systems (“smart” devices) ensures that the volume of information exchanged between virtual and physical worlds will continue to multiply, offering ever more data for machine learning models to use to learn and improve, faster and faster.

It’s easy to see the potential these tools have to change how we do many things, and some have even begun to wonder if we’ve created new sentient beings. But few of these conversations address how this new form of intelligence interacts with instinct, intuition, and feelings, and what this will mean for both humans and machines through the lens of evolution. After all, evolution never progresses toward a specific end destination—it merely adapts to changing circumstances. As human intelligence has changed, so have the roles of instinct and intuition, going far beyond simply helping keep us alive to informing the development of storytelling and science, math and mysticism, poetry and philosophy.

If we want to a clearer view of how these new technologies might interact with the evolution of life on Earth, we have to approach these ideas interrogatively. Thus, SMALL V01CE is an exhibition full of questions. What does the rise of large generative models mean for human instinct, intuition, and feelings? Will these tools enhance or dampen humanity’s innate instinct, as well as the processes by which intuition is refined? Will machines be able to observe, quantify, and classify forms of human instinct and intuition in ways we currently cannot? In not operating as fully rational agents, individual people often defy exact prediction—would more refined models of intuition change that? We often herald creativity as a quintessentially human endeavor—if its role is changing, what will this mean for the creative process and the production of art? And what might such pursuits mean for the development of new generative engines? Is it possible that machines will develop their own forms of instinct and intuition? If they do, would we be able to recognize them? And what would that mean for the future of creative expression?

Underlying these questions are considerations around first principles: are instinct and intuition productive aspects of intelligence, or evolutionary byproducts human beings have exapted? Likewise, are instinct and intuition critical for the production of art? Would a future intelligence capable of creativity need them in order to produce meaningful art? How might machine intelligences interact with non-human biological entities—be they bacteria, plant, animal? And zooming further out: what does it mean that we are conducting this experiment on ourselves at a time when we are still grappling with legacies of colonialism and oppression, with belief systems that foreground competition, extraction, and aggression? Why are we subjecting ourselves to this experiment, putting ourselves at risk in unpredictable ways?

Such questions don’t currently have clear answers and maybe never will. SMALL V01CE invites leading artists whose work engages these questions—artists whose interactions with these technologies are rooted in unconventional modes of knowing and perceiving—to share their own hypotheses, questions, reflections, and portals.

Exhibiting artists: Memo Akten, Minne Atairu, Nolan Oswald Dennis, Lins Derry, Linda Dounia Rebeiz, Behnaz Farahi, Holly Herndon & Mat Dryhurst, Lauren Lee McCarthy, Sara Ludy, Parag K. Mital, New Mystics*, Alexander Reben, Reeps100 & Trung Bao, Landon Ross, Rachel Rossin, Caroline Sinders, and Kira Xonorika. 

*New Mystics, organized by Alice Bucknell, features the work of Rebecca Allen, Zach Blas, Ian Cheng, CROSSLUCID, Patricia Domínguez, Dorota Gawęda and Eglė Kulbokaitė, Sadia Pineda Hameed & Beau W Beakhouse, Joey Holder, Evan Ifekoya, Bones Tan Jones, Lawrence Lek, Haroon Mirza, Tabita Rezaire, Tai Shani, Himali Singh Soin, Jenna Sutela, Saya Woolfalk, and Zadie Xa. 

[Quick update before I dive in: I’ve recently launched the Reality Studies podcast. If you’re subscribed here, you’ll see these episodes in your inbox, and/or wherever else you listen to podcasts. But if you prefer seeing the full videos, Reality Studies is also on YouTube! Here’s the first episode with Taylor Lorenz.]

One of the less-hyped aspects of the generative AI discourse is how it will shift our relationship to genre—and genre’s relationship to us. Though genre might initially seem like a secondary consideration in creative production, it plays a critical role in establishing audiences’ expectations and in the ultimate reception of artworks and creative acts.

The imagined ideal of the lone artist toiling away in the privacy of her studio, away from pesky considerations of audience, critics, and market, is already understood to be fallacy in the contemporary creative environment—which, like every other industry or pursuit, is deeply mediatized. So the idea that an artist or designer wouldn’t in some way consider genre or creative canon in their own outputs—or that it might *gasp* inform how they craft their work—feels silly at this point. But where I think generative AI stands to further trouble this boundary is through the centering of prompting as a key interaction mechanism with generative engines. My sense is that, as the practice of prompting continues to emerge as its own craft, the more pronounced genre’s role will become in the creation of art and design.

This evolving relationship to genre points to a future with new hybrid creative forms, but it gets even weirder to imagine how the respective conventions and boundaries of genre will be historicized, (re)imagined, and (re)incorporated into creative acts. In other words: a feedback loop is created when human creators deepen and complicate genre by metabolizing these outputs and then re-engage with generative tools. Set, repeat.

Check out the Reality Studies podcast!

What follows are a few case studies through which we can evaluate how genre might evolve in the wake of publicly accessible generative AI tools, and how that could impact future creative efforts. I hope to demonstrate how they embody a metamodern mode of cultural production and ultimately reinforce arguments I have made about the Postreality paradigm.

You don’t need to know too much about either Postreality or Metamodernism to read this essay. I will explain my sense of how generative AI intersects with these terms in greater detail in the conclusion, but by way of a quick foundation:

• Postreality is the reality paradigm that emerged after (and from within, and in contradistinction to) Modernity, bubbling up in the wake of the Second World War and persisting through today. It is typified by non-monolithic, interdependent, synthetic, and more-than-human approaches to the social construction of reality (establishing consensus with each other about what is real). More on Postreality here.

• Metamodernism is described as a structure of feeling that arose after postmodernism, which describes an “oscillation” between the aspects of modernism (yay grand narratives, individual genius, and universal ideals!) and postmodernism (everything is relative and grand narratives are dangerous; irony is a scalpel). Since the resurgence of the term in 2009, scholarship across different fields has broadened and diversified its meaning and usage (competing opinions proliferate). If this subject appeals to you I recommend this piece for a crash course in recent developments in metamodernism/metamodernity and how they do and don’t fit together. For my purposes in this essay, what I see as key to understanding metamodern creativity is a reconstitutive, hybrid approach to making—one that is in constant dialogue with other social, creative, and artistic content as a function of online interconnectedness.

Zombie Wes Andersonism

Earlier this summer, the “Wes Anderson” trend—in which users presented their lives ostensibly “in the style” of the film director—swept TikTok. Then, some began to use image generators to riff on the trend, assembling trailers for films in other major franchises including Harry Potter, Lord of the Rings, and Star Wars—as directed by “Wes Anderson.”

Wes Anderson is riffable in this way because his style is idiosyncratic, and has maintained some degree of aesthetic continuity across his nearly 30-year career. If you’ve been on the Internet, you have an idea what this entails—the tracksuits, subjects framed in the center, the subtle interferences with the fourth wall, the earth tones and playful pastels, the twee vibes.

But how connected are these techniques to the experience of watching an actual film by Wes Anderson, like his latest, Asteroid City? What if these are actually reinforcing a simulacrum of Wes Anderson that’s not really accurate? That’s exactly the claim Stuart Heritage recently made in The Guardian:

The problem is, though, that none of these things actually spoof Wes Anderson. The only thing they do is spoof other Wes Anderson spoofs, in particular Saturday Night Live’s Wes Anderson horror movie spoof from four years ago. Like all the newer parodies, they were just a loose collection of tropes – kids in matching tracksuits, old tents, the colour brown – except it was made by incredibly talented people with a large production budget and a clear appreciation for Anderson’s work, as opposed to a computer program that can stitch Bill Murray’s face on to the body of Gandalf so ineptly that it looks like every nightmare you’ve ever had rolled into one.

Read more

An earlier version of this article appeared in Forbes.

Generative artificial intelligence, a loose category to describe a variety of machine learning techniques that produce text, images, video, and audio, and other content, has provoked heated debate about the relationship between humans and machines.

Conversations about AI safety, ethics, biased data, a…

Read more

The “Critical Readings in AI” series is focused on helping develop informed approaches to AI. Each installment synthesizes resources that have helped me clarify my own position and thoughts on what’s at stake, as well as situate them within the bigger picture. Resources included here are not endorsements; they are here because they provide a frame on the subject that I think is useful to understand, not because I intend to promote anything or anyone outright.

In the first installment of this series, we took a look at some of the labor questions that the rise of generative AI provoked. For this volume, we look at another contested topic: the future of health and medicine with generative tools.

It’s no surprise that medicine is often referenced by proponents as a leading use case for generative AI. On its face, it’s hard to argue with the idea that we’ll be able to accelerate critical medical research, formulate novel treatments, decrease costs, and improve access to positive outcomes around the world.

Of course, it’s not quite so simple—with these new opportunities there are new risks, ranging from over-reliance on these tools by medical practitioners, to skills gaps using these technologies increasing inequality, to the ways these tools will be implemented to cut costs in hospitals, to bad actors using these tools to develop harmful biological formulations (i.e., chemical weapons). These opportunities and risks are outlined in what follows.

Read more

It’s been almost two months since Apple announced its long-awaited Vision Pro headset. I’ve read the takes, I’ve talked to many different folks, including friends and colleagues who worked directly on the device. There’s so much to be excited by in the headset, but I can’t shake a gnawing sense of concern about it.

I remember when I first got started connecting with folks in the virtual reality (VR) community in 2014, I would encounter industry veterans who grouchily waved off the grandiose claims about VR during the phase of renewed interest following the Oculus kickstarter in 2012. They had been through too many headfakes to believe it was really happening this time—they’d heard it all before, they’d even made the same points about immersion as were being made now—and what had any of it amounted to? As someone new to the scene, it had a distinct “turd in the punchbowl” flavor; my intention is not to present as a kneejerk crank now.

That said, with nearly a decade of experience, I know much more about immersive technologies, and I too have developed stronger critical positions about how, why, and when we should be using them. I have more empathy for their position, but I also want to honor the excitement brewing around the Vision Pro. To that end, some quick groundwork:

• Generally speaking I think the announcement of the Vision Pro is a Good Thing™ for the XR industry (inclusive here of virtual, augmented and mixed reality; spatial computing; adjacent technologies like haptics and smart devices; and the still-contested notion of the metaverse), to which I’ve dedicated nearly a decade of my attention and interest.

• Those who have positively reviewed their experience of the headset are longtime industry journalists whose voices I trust.

• Apple has a proven track record of identifying the “right” time to enter (and “win”) a given market, and is a strong enough center of gravity that social practices have morphed around their products in the past (see: the social acceptability of walking around with AirPods in your ears).

• Yes, the device is expensive enough to price out the vast majority of potential consumers worldwide. No, this is not that surprising. Yes, Apple has released equivalently costly items that succeeded in the past. Some people even think should be more expensive.

• There is dissonance between how the device was marketed (fun entertainment and workplace applications!) and who it is for (developers and early adopters), at least until it comes down in cost or new models roll out that are more capable (or have similar capabilities at cheaper costs, a la iPhone SE). This type of delta between expectations created through marketing and actual product capabilities has dogged other big names in XR (see: Google Glass, Magic Leap whale, Meta’s “metaverse” offerings). In Apple’s case, that’s less a matter of using VFX to present impossible XR experiences and more a matter of having a robust enough content/app ecosystem and large enough user base to make the device desirable to all but a tight core of niche users and developers.

My Concerns with Vision Pro

In the weeks since the announcement, two considerations have emerged as the primary sources of concern regarding the Vision Pro. One has to do with Apple’s messaging, which is then connected to a broader narrative that Apple fits within.

Read more

Sometimes it really is wild to confront how rapidly individual developments in science and technology can balloon into world-altering innovations. One such development: the transformer, the deep learning network architecture upon which contemporary large language models (sometimes also called “foundation models”) like OpenAI’s GPT, DeepMind’s AlphaFold, and Google’s LaMDA are built.

Last Monday, the Transformer celebrated its 6th birthday.

What is a transformer model? A plain-English description from the Nvidia blog:

A transformer model is a neural network that learns context and thus meaning by tracking relationships in sequential data like the words in this sentence.

Transformer models apply an evolving set of mathematical techniques, called attention or self-attention, to detect subtle ways even distant data elements in a series influence and depend on each other.

As AI Scientist Dr. Jim Fan explains, the “Transformer did not invent attention, but pushed it to the extreme.” (That honor goes to “Neural Machine Translation by Jointly Learning to Align and Translate,” which Yoshua Bengio’s lab published in 2014, and in machine learning circles is regarded as a major milestone in natural language processing. Bengio has made headlines recently for joining fellow AI OG Geoffrey Hinton in adopting an alarmist stance, citing the existential risk AI poses to human civilization).

Transformers “push [attention] to the extreme” in their introduction of so-called “attention mechanisms,” which make it possible to track the connections among words, forward and backward, across even long texts. Coupled with their ability to process data in parallel rather than sequentially, they can map these connections faster and with more efficacy than prior deep learning models.

I’m not an AI engineer and won’t pretend to be—if you want to dive a bit deeper into transformers, I recommend this piece in TechTalks and the aforementioned piece on the Nvidia blog.

In the course of my research developing the manuscript for Postreality (more context for Postreality here), I’ve spent a lot of time in the weeds with the many and varied theories of consciousness. One that has been fruitful in my own thinking is the Attention Schema Theory (AST), developed by neuroscientist Michael S.A. Graziano’s lab. AST is described as follows in its proposal:

The theory begins with attention, the process by which signals compete for the brain's limited computing resources. This internal signal competition is partly under a bottom-up influence and partly under top-down control. We propose that the top-down control of attention is improved when the brain has access to a simplified model of attention itself. The brain therefore constructs a schematic model of the process of attention, the ‘attention schema,’ in much the same way that it constructs a schematic model of the body, the ‘body schema.’ The content of this internal model leads a brain to conclude that it has a subjective experience.

I encountered the theory through Graziano’s 2016 essay in The Atlantic, and later through his book Rethinking Consciousness. I highly recommend reading at least The Atlantic piece, which charts the evolution of sensing mechanisms in animals:

The theory suggests that consciousness arises as a solution to one of the most fundamental problems facing any nervous system: Too much information constantly flows in to be fully processed. The brain evolved increasingly sophisticated mechanisms for deeply processing a few select signals at the expense of others, and in the AST, consciousness is the ultimate result of that evolutionary sequence. If the theory is right—and that has yet to be determined—then consciousness evolved gradually over the past half billion years and is present in a range of vertebrate species.

It’s possible this theory won’t prove true with regard to the emergence of consciousness (which remains a subject of fervent debate), but it’s hard to argue that viewing the development of our own brains and nervous systems through the logic of selective signal enhancement (the process by which certain sensing systems improve for purposes of evolutionary fitness) is a helpful way of grasping our sensory experience of reality. Moreover, if there is truth in it, it could hold important implications for the future of the humble transformer and AI more broadly.

It’s easy to get caught up in reductive semantic comparisons and wander away from anything useful; of course the “attention” in AST is a different formulation than the mathematical function and weights in a transformer. I want to be extremely careful in anthropomorphizing machine intelligence, as that kind of thinking is prone to category errors. Still, I can’t shake the sense that we’re getting at something critical for understanding the future of machine intelligence. The breakthrough of transformers was pushing attention to the extreme—in other words, granting machines enhanced senses for mapping relations across a given dataset and filtering for relevance.

Current evidence says machines do not have subjective awareness, and many argue that that is an impossibility, but ongoing discussions of existential risk posed by AI lead many to wonder what would happen if machines did develop consciousness. What would that look like? What would they think and desire? What would they ask us? Would we even know it happened? If so, how?

The buzziness around this conversation is an entrypoint into the broader conversation about complex systems, which are characterized by individual parts that interact with each other in a variety of different ways, leading to randomness and emergence. According to Sante Fe Institute Co-Founder-in-Residence David Pines, emergence “refers to collective phenomena or behaviors in complex adaptive systems that are not present in their individual parts.” One way of thinking about emergence is as the aspect that makes the whole greater than the sum of the parts. Further, emergence can describe properties in systems that are in fact properties of the system’s relationship to the larger environment it is part of, such as the human immune system interacting with external pathogens.

With companies, universities, and nations all in an arms race to develop increasingly powerful AI systems, it’s safe to say a complex system has formed (and continues to evolve) around generative AI—and machine consciousness is only one of many possible properties that could conceivably emerge as a result. In fact, what emerges is more likely to be something we don’t predict, or to diverge from expectations in ways that only make sense in retrospect. As Scott J. Shapiro recently pointed out: “Predicting the future is difficult because the future must make sense in the present. It rarely does.”

Making the case for AST, Graziano describes how the evolution of the tectum during the Cambrian explosion ~500M years ago gave vertebrates a centralized controller for attention that could coordinate among senses, which formed the basis of what would one day become the cerebral cortex in humans, a development he believes was critical for the emergence of what we now call consciousness. Under this view, consciousness and other critical aspects of modern human beings can be seen to have emerged through the long and often arbitrary nature of evolution. In machine learning, human beings are innovating using the vastly more time-efficient scientific method, cranked up to an 11 through the incentives of late capitalism. Who knows what will emerge, but we would be wise to view innovations like the “attention mechanisms” of transformers as targeted cognitive enhancements that facilitate emergence.

Transformers, only six years old—and only really in the public eye for half that time—have already activated the energies of folks all over the world. And it’s exceedingly possible that a new proposal will come along that will introduce another order-of-magnitude evolution in the next few years. With so much commotion, would we even notice if a truly novel property emerges in a given model? What the explosion of energy around transformers tells me is that our best bet will be to keep paying attention to attention.

Last week, and I wanted to spin up a space to live react to the announcement for (what we now know is) the Apple Vision Pro. In evaluating our options, we landed on a Twitter Space; even in the face of Twitter’s ongoing enshittification, it was still best-suited to our needs.

That morning, it occurred to me that Substack Notes could (and should) offer a similar service, which ultimately led me to the idea that Substack and Clubhouse should merge.

For a few months there in Spring ’21, Clubhouse was the place to be. It raised $100M on the belief that it was an exciting new angle for the future of social media. In 2023, the situation looks decidedly less optimistic for the company, which recently had to lay off half its workforce. We can point to a number of external factors—hype fading, vaccines, less stringent quarantine protocols, the rollout of Twitter Spaces—but it’s not because Clubhouse’s product doesn’t work properly. In fact, speaking purely from anecdotal experience, I find Clubhouse tends to work better than Twitter Spaces. It just doesn’t have the broader appeal and community of users that Twitter has.

Meanwhile Substack, still ascendant, has rolled out Notes, a well-intentioned stab at social media that has nowhere near the gravitational pull of Twitter, and still lacks major features which would cast it as anything close to a Twitter competitor. To be clear: I don’t believe Notes needs to be a full-fledged Twitter competitor—and attempting to do so could lead to the replication of perverse incentives that have enshittified other social media platforms. But I’d be thrilled to see Notes become a thriving (yet still niche!) space that fosters knowledge sharing, civil debate, and critical thinking. For that to happen, it will need better UX and more people who want to use Notes as audience-members and commenters outside of specific newsletter siloes.

A merger of Substack and Clubhouse could not only address their respective obstacles, but open up possibilities for fostering new forms of prosocial interaction online. It could also offer a new vision for a social media revenue model not dependent on advertising—something I trust many people would embrace.

So this is my humble submission to the ideasphere in the hopes that one day I can spin up Substack Houses for all my social audio needs.

Share

Generative AI is a subject that has prompted renewed talk of the Singularity, the supposed moment when an “artificial superintelligence” explodes past human capability, displacing us as the dominant species on Earth.

There are very real concerns associated with the rise of publicly accessible generative tools, but these have to do with the ways they will extend, exacerbate, and accelerate existing systemic problems. As I wrote in my recent piece on the “Creative Singularity”:

[M]any of the zealous proclamations from prominent figures in the field feel divorced from reality. Furthermore, they bypass less extreme but more applicable lessons from the past that must be reiterated in the face of hype, which tends to erase guiding insights.

Subscribe now

I empathize with the fact that the subject feels overwhelming, and these doomsday visions play into the Frankensteinian “original sin” of sci-fi: that through our hubris we create something we cannot control, and thus spells our doom. But the framing of AI as an apocalyptic entity that will overthrow humanity is unproductive at a time when literacy is critical for developing practical solutions and frameworks.

What follows is a new series focused on helping develop informed, critical approaches to AI. Each installment synthesizes resources that have helped me clarify my own position and thoughts on what’s at stake, as well as situate them within the bigger picture. I want to say from the outset that this series will include information and works I both agree and disagree with; they’re here because they provide a frame on the subject that I think is useful to understand, not because I intend to endorse anything or anyone outright.

For the first installment, we take a look at the labor questions that are emerging as a result of generative engines. It’s a big, thorny subject, and one that needs a clear context established amid a lot of noise.

Starting broad: how exactly will generative tools be used to extend and accelerate the existing labor paradigm? And what problems will be replicated or intensified as a result? In the following essay, Ted Chiang argues that generative AI will be used to consolidate power among the wealthy at the expense of laborers by offsetting decision-making and accountability, just as management consulting firms did over the past few decades:

Will A.I. Become the New McKinsey?

As it’s currently imagined, the technology promises to concentrate wealth and disempower workers. Is an alternative possible?

Ted Chiang

Is there a way for A.I. to do something other than sharpen the knife blade of capitalism? Just to be clear, when I refer to capitalism, I'm not talking about the exchange of goods or services for prices determined by a market, which is a property of many economic systems. When I refer to capitalism, I'm talking about a specific relationship between capital and labor, in which private individuals who have money are able to profit off the effort of others. So, in the context of this discussion, whenever I criticize capitalism, I'm not criticizing the idea of selling things; I'm criticizing the idea that people who have lots of money get to wield power over people who actually work. And, more specifically, I'm criticizing the ever-growing concentration of wealth among an ever-smaller number of people, which may or may not be an intrinsic property of capitalism but which absolutely characterizes capitalism as it is practiced today.

Full article here.

It’s also important to rewind a bit and understand the labor context in which these tools were created, which followed existing lines of exploitation under global capitalism. This piece also makes clear why “old school” labor solidarity and organizing will be critical in ensuring human wellbeing throughout the transition toward deeper automation:

Read more

[NOTE: Join Jessica Brillhart and me for live takes on Twitter Spaces during & after the WWDC keynote, kicking off at 10 am PST June 5. We’ll be talking shop, sharing takes, and hanging out with old and new friends.]

Though it’s still not official, all signs point to the fact that—after years of development, leaks, and no shortage of internal drama—Apple is at last (allegedly!) wading into the XR headset waters, based on heavy speculation that they will make an official announcement during next week’s WWDC conference (this year’s tagline is “Code new worlds” and includes an augmented reality easter egg).

The alleged details: it will be a device called Apple Reality Pro, rely on an external battery pack, and cost ~$3,000. It will be a “mixed reality” headset with both virtual reality and augmented reality capabilities (the latter through what’s called “pass-through” AR, which involves representing the physical world in the headset using cameras/sensors on the device).

I’ve been participating and working in the XR industry in various capacities for nearly a decade, and Apple’s entry into the arena has long been heralded as the moment that VR and AR will truly go mainstream, reprising the company’s glorious “iPhone moment.”

Subscribe now

Candidly, I’m excited to see what they have on offer, but I see this as mostly a symbolic entrance—especially if we’re to believe a recent report that expected production for 2023 is only 100,000 headsets. This is not the moment for mainstream XR adoption; this is the moment for Apple to get developers excited to build apps and content for this headset—and that’s a perfectly reasonable strategy.

I’ve been predicting for a while that headsets won’t follow a mobile device growth curve for the simple reason that they don’t solve immediate problems the way that mobile phones with Internet capabilities did, and to boot involve a whole lot more friction (they decidedly do not fit in your pocket). They open up new ways to work, play, and interact that will be appealing in the long term, but I see this as a long, more linear growth process. And out-of-the-gate, Apple’s really only trying to get in front of power users and tech adopters with excess capital, already quite niche. One of Apple’s longstanding superpowers as a company has been its ability to present a vision for the future of creativity to creative professionals through computing. If they play their XR cards right, I think this is something that is wide open for them to repeat—especially after Meta’s attempts have fallen flat over the past couple years.

What I think is most at stake for Apple next Monday is actually conceptual.

Read more

It seems like some unspoken rule of the Internet: every few years, the “[X] is Dead” formula must be activated. This go-around it’s the metaverse in the crosshairs. I probably shouldn’t take the bait, but I can’t help myself.

No, the metaverse isn’t dead. I don’t say that as some grand gesture of support or rallying cry; how the metaverse fares as a marketing term or investment category are only surface reflections of a deeper discussion about the future of connectedness and embodiment in digital contexts.

My problem with critiques like Ed Zitron’s are not that I disagree with the arguments he makes on the page. In large part I agree with him: Mark Zuckerberg’s coopting of the term was (charitably) short-sighted bluster and (uncharitably) expedient for near-term shareholder interests at an otherwise fraught moment for the company. The blind frenzy that ensued across other tech giants, venture capital, and media portrayals only amplified the disorienting blur through which the term was catapulted into the public eye. And for all intents and purposes, it does appear that Meta has de-emphasized the metaverse as a marketing angle in favor of the buzzier “generative AI”—though my spidey sense says it’s more of a hibernation while necessary technologies and social fluencies catch up before a return to beating the metaverse drum.

What I find frustrating about Zitron’s argument is that he conflates the metaverse as a term with the set of ideas, practices, and developments it represents. Put another way: we might not use the term “cyberspace” much anymore, but you’d never say, “cyberspace is dead.” At one point, “cyberspace” was the container for discourse around virtual life, and it facilitated developments that continue to impact our experience online today.

Over time, its function as a driver of fresh ideas faded. Since then, different ideas and terms took center stage (remember “transmedia”?). In the early 2020s, the metaverse took the mantle in a loud way, starting with Web3 folks on Clubhouse in the stir-crazy spring of 2021 (yes, Clubhouse is the first place the metaverse conversation kicked off), mounting throughout the year until the crescendo of Facebook rebranding to Meta.

I metaphorize the metaverse as the eye of a storm: at the center there’s no consensus definition, but the storm is real. My sense is that the metaverse will continue to be the container for an important set of ideas and dialogues, but even if it’s not, they will persist through other terms. And how these develop have material impacts on all of us. In his recent book Meganets, David Auerbach puts it as such:

For all the talk of the metaverse—much of it merely a debate over what the metaverse even is—the fundamental difference between the metaverse (whatever it will be) and the internet today is not quality but quantity. The metaverse threatens to add virtual reality (fake physical space), “augmented” reality (digitally enhanced real physical space), and cryptocurrency-driven NFTs (non-fungible tokens, which are artificially scarce digital assets to buy) to the existing potent online world, but all of the metaverse's changes will amount to the same thing: more data, more connectivity, more transactions, more complexity. The movement from Web 1.0 to Web 2.0 to the metaverse and onward has not stemmed from radical new technologies. Rather, the change in online life is like the evolution of the brain, growing ever more complex until surprising new phenomena emerge purely through that complexity—along with that much less ability to understand what is going on. The metaverse, for all practical purposes, is already here, growing incessantly. We lose more control to it by the day.

I understand the impulse to dunk on Zuckerberg and Meta. I worked for many years as a journalist covering XR—including predicting Facebook’s metaverse play back in 2016—and I know how irritating it is to be inundated with hype about the “next big thing.” And of course there’s a righteous thrill in calling out the powerful for their misdeeds.

But taking such a narrow view of the metaverse does a disservice to audiences. Technology develops on long, messy timelines, full of false starts and transient terminology. But technology is a major vector for power in contemporary life, and conditioning audiences to laugh off arenas poised to impact their lives, rather than critically grapple with the underlying ideas, is at best a waste of time.

In writing the futures reports PORTALS and Simulating the Future, Leah Zaidi and I tried to take a measured but expansive approach to how the metaverse would impact our lives in the future, using her Strategic Worldbuilding methodology. Our intention was to engage critically with the topic without endorsing the ways the term was being wielded for hype.

I bring it up here as a shortcut to some of the counterarguments I’d make to the notion that the metaverse is dead. Yes, a lot of the developments related to the metaverse post-Meta rebrand have been cringey. But we’re fools if we let that stupefy us into disregarding all the social, cultural, economic, and even political energy that was activated over the world as a result. Better that we acknowledge that it will determine the shape of our digital lives for years to come, and try to use whatever agency we have to influence it toward humane outcomes.

Share

Over the weekend—April 30th, auspiciously enough—the world wide web celebrated 30 years of launching into the public domain. Though Tim Berners-Lee wrote the first proposal for the web in 1989, and it was available to the public in 1991, it wasn’t until 1993 that the web was released to the public without any patents or fees.

It’s easy to conflate the Internet and the web, so to make sure we’re starting on the same page (from the NPR piece):

[The Internet] was also difficult to use. To read a story from NPR, for example, you would need to know which network-equipped computer had the file you wanted, then coax your machine into communicating directly with the host. And good luck if the computers were made by different manufacturers.

But 30 years ago this week, that all changed. On April 30, 1993, something called the World Wide Web launched into the public domain.

The web made it simple for anyone to navigate the internet. All users had to do was launch a new program called a "browser," type in a URL and hit return.

In this way, the web rendered the Internet more accessible—and opened up to new modes of expression, connection, and exchange. And because Berners-Lee convinced CERN to release it into the public domain, it became a public good that was able to develop freely.

A few years ago, the notion of “Web3” began to surge in popularity. Proponents argue the web has undergone three major phases: the original web of the early- to mid-1990s (which broadened information access to the many), Web 2.0 (which emphasized user-generated content and social participation), and now Web3 (which emphasizes decentralization, openness, and data sovereignty, and typically involves the use of blockchain technologies). Notably, Berners-Lee once proposed his own version of “Web 3.0” in 2009, which overlaps in some ways with the current framing in foregrounding data ownership and privacy, but doesn’t rely on blockchains, of which he remains skeptical.

It’s a weird time for Web3. Sure, we can reference the dismal state of affairs for NFTs—but I’ve lived through enough boom-bust cycles to know it’s prudent to be as cautious dancing on graves as you are surfing on hype. Practically speaking, in the ~3 years since Web3 notionally caught steam, and was heralded to change everything, very little about how we interact with the web has actually shifted from Web 2.0. It’s still more a set of ideals and experiments rather than a meaningful “disruption” to the ways power has consolidated among a handful of major players online. The crypto winter is giving cover for “buidlers” to focus on developing solutions that might not make for grabby headlines, but push the field further in the direction of being actually useful. I’m intrigued, for instance, by the prospects of the ERC-6551 standard, through which every NFT could have its own wallet.

Where I remain most excited are the possibilities of Web3 to extend the legacy of the early web through new infrastructures for public goods and commons and to use our understandings of what’s harmful about Web 2.0 to drive new understandings of individual and collective sovereignty. I encounter these efforts in my work with Protocol Labs (Disclosure: I earn income with them and hold FIL), through which I found the Funding the Commons event series and Kevin Owocki’s Green Pill podcast (both of which are great resources to find folks thinking deeply and less hype-ily about this stuff).

Last fall, I co-developed the Web3: How It’s Going summit with Zak Robinson, Nancy Baker Cahill, and Vellum LA. I keep coming back to something that Starling Lab Founder Jonathan Dotan said in his keynote, which outlined how the research center is using blockchain technologies as vehicles to securely document war crimes in Ukraine and elsewhere around the world. He said:

We need to ensure that this type of technology, which is doing this form of authentication, can’t run on hype. It’s something that requires much more than money. I want to suggest to you that in this new era that we’re entering with Web3, that there might be a different origin story for the purpose of what we’re doing. Although a lot of the technology began with work that was done during the financial crisis, trying to provide an alternative for the banking system, I wonder, maybe this is the moment where we start a new chapter and we think about the integrity of the Internet and what is under siege. Because let me be clear: Russia is in the process of removing itself from the Western Internet, so the type of authenticity that we are creating here is vital to the survival of the global Internet—it is vital to the integrity of the internet as we know it—and I want to end by saying that, although there are meaningful criticisms of Web3, and that there are reasons to believe we in fact have lost our way in thinking about the type of greed and incentives that might be animating this space, I’d argue to you that we shouldn’t run away from this type of criticism.

Instead, these are the criticisms we should lean into. The criticisms are guides for how Web3 can lean into the unique affordances of blockchains to actually make the web better.

The world wide web was novel because it created a useable interface for people to participate on the Internet, and because of the decisions of its founders, it did so as a public good. The key proposed contribution of Web3 is decentralization—the prospect of automating and scaling trust, transparency, and governance. Like Jonathan, I still see a lot of flaws and perverse incentives built into the current shape of Web3. The crazy speculation of the NFT boom made headlines, and attracted a lot of people before the technology and the existing industry was quite ready for primetime. And hey, maybe the tech will prove to be as useless as the detractors say it is (though with the flood of unverifiable content coming our way via generative AI, I doubt it). Until such a time, I see the technology as the embodiment of a set of ideals—however flawed or compromised—about what the web could be. I think a lot of people generally agree with these ideals; it would be great to experience an Internet where we could choose how our data was used and for what purposes, which foregrounded provenance, authenticity, and verifiability.

That’s why I still see a lot of possibility in the promise of Web3, and why I think it’s premature to throw the baby out with the bathwater—though the grifters and culture of greed are undeniably frustrating. Ultimately, the technology and the community surrounding it are imperfect and still relatively nascent. I’m excited by those who continue to use this moment as an occasion to (try to) produce more equitable outcomes, improve access to knowledge and information, and scale decisionmaking structures that benefit the good of the many over the interests of a select few.

When our efforts are directed toward embracing these possibilities—whether or not that involves the word “blockchain”—we carry the legacy and core ideals of the original web.

Share

⏰ Reading time: 2 minutes

Subscribe for more like this by clicking below:

Subscribe now

This week, we look at the real impacts of fascist communities that form in virtu…

Read more