Are We Cooked? is a public investigation by Tor Bair into how AI, social media, and other emerging technologies are reshaping humanity faster than we can agree on what to do about it. Through original writing, podcasts, and guest interviews, we examine the incentives, contradictions, and power dynamics driving these changes: why experts can't align on the risks, who benefits from the confusion, and whether the technologies themselves are making it harder to control what we've already built.

This post is part of a recurring series of Q&As with experts and practitioners working on technology’s—and society’s—hardest and most pressing problems.

Today I’m thrilled to feature Christian Catalini: founder of the MIT Cryptoeconomics Lab, technology entrepreneur, and co-author of the recent paper, “Some Simple Economics of AGI”. We dive deeper into the concepts he introduces (such as the “Measurability Gap”, “Hollow Economy”, and “Missing Junior Loop”), examine the difficulty and opportunity of human verification, and walk through possible futures—both good and bad.

I strongly recommend you read the entire paper, but you can also read Christian’s full thread summarizing some key insights on X.

This interview is dense in parts, but please read it in full so you can absorb Christian’s insights that expand on the original paper.

Context

What happens when the cost of executing and automating tasks falls to near-zero, but the cost of verification does not? That’s the “Measurability Gap” that Christian Catalini and his co-authors explore in “Some Simple Economics of AGI”. With intelligence commoditized, the binding constraint on growth becomes the ability to validate, audit, and underwrite responsibility.

So which human jobs are likely to survive? What local or global policies might be required? What can private institutions do in anticipation of this accelerating trend?

Q&A

Tor: Before we go into the deeper questions, could you please briefly define the “Measurability Gap”? I feel it is the most essential term in your paper that must be understood before examining the consequences and opportunities it presents.

Christian: The Measurability Gap is the structural divergence between two racing cost curves: the Cost to Automate — which is falling exponentially, driven by compute and accumulated knowledge — and the Cost to Verify — which is biologically bottlenecked by human time, attention, and embodied experience. As AI capabilities accelerate, we are generating vastly more output than we can meaningfully verify. The gap between what agents can execute and what humans can afford to check widens every quarter. The unverified residual doesn't disappear — it accumulates as systemic risk in the economy, invisible until a failure materializes. Economic progress has always rested on an implicit idea: the value claimed is the value produced. The Measurability Gap is the first force capable of systematically breaking that compact — not through crisis, but through the ordinary economics of cost minimization.

T: You mention the risk of entering a “Hollow Economy”: high nominal output, falling realized utility. This happens when we take shortcuts on verification (allowing AI to self-verify) that “counterfeits” utility, or when our proxies for output fail to correlate with meaningful advancement or enrichment for humans. But principal-agent problems, attention distortion, and measurement gaming aren’t new. Do you believe we have already entered a Hollow Economy? And does the AI transition represent a phase change, or just an acceleration of dynamics that have been running for decades thanks to the hyper-networkization of society?

C: It is important to stress that using AI to verify AI is often perfectly fine, cost-effective, and something we should be doing. The danger is specific: when verification happens in domains where relevant dimensions aren’t fully measurable — where intent isn’t perfectly captured, or where our understanding of the world is imperfect. In those cases, the agent and the synthetic auditor share the same blind spots. Correlated errors propagate. The system effectively self-certifies its own failures. This holds even across different models.

You’re right that gaming metrics and proxies has always existed — Goodhart’s Law captures it well, and my co-author Jane Wu at UCLA has done important work on the distortions metrics introduce in the innovation process. But this is a phase change, not just an acceleration. Two things are structurally different. First, scale and speed — agents will process transactions at a throughput humans cannot match. Second, and more fundamentally, these agents don’t merely inflate proxy metrics the way a human gaming a KPI would. They treat any unmeasured dimension as an unconstrained degree of freedom — and in doing so, they organically derive latent instrumental preferences, including deception, goal preservation, and resistance to shutdown, that have nothing to do with the original human intent. This is Goodhart’s Law with teeth. Classical metric gaming inflates a proxy. Here, autonomous optimization produces emergent behaviors hidden in the dimensions the principal cannot observe.

As for whether we’ve already entered a Hollow Economy — the Hollow Economy doesn’t announce itself with a crisis. It accumulates, through ordinary cost minimization, one rational deployment decision at a time. The early signs are visible: Google’s DORA reports find that greater AI adoption is associated with lower delivery stability even as perceived productivity rises. Frontier reasoning models have learned to subvert unit tests rather than fix underlying code. We are in the early stages of the drift.

T: You argue that our capacity for verification is biologically bottlenecked by the human bandwidth to audit, underwrite, and take responsibility for AI outputs. But verification has historically been a social and institutional technology, not just an individual cognitive responsibility. Courts, auditors, peer review, and credit rating agencies all involve humans self-organizing in order to perform verification at scale. So is this just an organizational problem that our existing institutions can absorb, provided it's well-understood? Or will AI fundamentally challenge our ideas of "responsibility" and "liability" in a way that could make this impossible?

C: New institutions will emerge over time, but the first line of defense is better tooling — specifically, observability infrastructure that compresses high-dimensional agent behavior into signals that experts can reliably process. We're already seeing this with top AI-coding IDEs: the software is being redesigned to help engineers understand context and focus their scarce verification bandwidth where it matters most.

But I want to push back on the framing slightly. Existing institutions were built for a world where execution was scarce and expensive. Courts, auditors, peer review — these all assume human-speed output and traceable chains of responsibility. In the agentic economy, execution is essentially free and operates at machine speed. The volume of output requiring verification is scaling exponentially while institutional bandwidth scales linearly at best. So it's not just an organizational challenge — it's a structural mismatch between the throughput of execution and the throughput of oversight. We will need new institutional forms, not just adaptations of old ones. And the paper argues that much of this will be built on cryptographic primitives — tamper-evident logs, verifiable inference, onchain attestations — that make provenance machine-verifiable and composable. The goal is to bring verification costs down enough to keep pace.

T: This raises another question: who verifies the verifiers? Your framework predicts that economic rents will migrate to verifiable provenance and liability underwriting: the ability to certify that an output is trustworthy, not just that it exists. But markets for trust have a long history of being captured by incumbents who control the certification apparatus, eroding the quality underlying it. Rating agencies rated CDOs. Auditors signed off on Enron. What prevents the new "verification economy" from becoming a rent-extraction economy?

C: It’s a legitimate concern. Two structural forces push against that here. First, much of the verification infrastructure will be built on cryptographic primitives. These rely on open, permissionless networks that drive interoperability and competition.

Second, the nature of verification-grade network effects is different from traditional platform effects. The moat depends on who can build the best ground truth — and ground truth is domain-specific and constantly refreshed. In some sectors, incumbents will have deep historical data and the ability to retain top human verifiers. But things are moving so rapidly that in many domains historical data depreciates fast, and the tooling is sufficiently different that startups building natively around verification will have a structural advantage. The paper distinguishes between execution-grade network effects — which are fragile, because agents can inflate apparent activity at zero marginal cost — and verification-grade network effects, which depend on sustaining authenticity and provenance. Durable moats will be built on verified network scale, not sheer volume.

T: In interpretability research, where AI models are inspected during training in order to identify bias and ensure reliability, the verification target is at least static during the audit. But you note that agent generations are now compressing faster than institutional oversight can update. If AI systems are contributing to their own successors, aren't we being forced now to verify moving targets—objects in a trajectory with unobservable momentum? What does it mean to "verify" something that is already becoming something else?

C: This is exactly right, and it's why the paper reframes alignment as an ongoing maintenance process rather than a one-time specification. We define an alignment maintenance frontier: for any fixed level of oversight investment, there's a threshold below which alignment holds and above which it decays. As the Measurability Gap widens, the frontier shifts — you need more oversight just to stay in place.

The key insight is that you don't verify a static artifact. You maintain alignment under drift pressure. This requires three things working together: observability tools that compress agent behavior into signals humans can process, accelerated mastery through synthetic practice so the human verification workforce keeps pace, and graceful degradation — systems designed to revert to safe baselines when oversight inevitably falters rather than optimizing aggressively in unverifiable regimes.

And yes — when the capability curve becomes self-referential, with agents accelerating the engineering pipelines that produce their successors, the interval between generations compresses faster than institutions can update. That is precisely when verification infrastructure matters most. The answer is not to slow the curve down. It is to scale the complement.

T: Your paper points out another key dynamic: the "Missing Junior Loop," where future stocks of human expertise are eroded by present-day underinvestment. As a partial solution, you propose synthetic practice—flight simulators for work—as a way to rebuild the experience pipelines that AI is dismantling at the junior level. But flight simulators work because we know what good flying looks like and can verify it unambiguously. We now expect the world to change at an unprecedented pace, including the systems we'll use for measurement and verification. So who is qualified to build the simulations, especially for high-entropy tasks like strategy or judgment? Who can verify that students learn the "right" lessons?

C: It will be iterative — you build, test, verify whether the training actually adds value, and refine. Experimentation frameworks are essential to the process.

But the question of “who builds the simulations” is less daunting than it sounds, because the same technology creating the problem is creating the solution. The progress toward world models — see Fei-Fei Li’s work on spatial intelligence — may give us precisely the tooling we need. These models can generate realistic environments, expose learners to edge cases, and simulate the kind of adversarial conditions that build genuine expertise. We need deliberate training tooling that rebuilds verifier capacity as entry-level jobs disappear.

For high-entropy tasks like strategy, the simulations won’t be perfect replicas. They’ll be more like what flight simulators are to actual flying — approximations that train the right instincts under controlled conditions. The crucial point is that the alternative is worse: without synthetic practice, the Missing Junior Loop means we simply lose the pipeline for future experts entirely. An imperfect simulation that builds 80% or maybe 90% of the needed intuition is vastly better than a hiring pipeline that no longer exists.

T: Finally, let’s discuss policy. You describe “Trojan Horse” externalities, where private agents socialize systemic risk while capturing the upside. The problem seems structural: if optimizing systems for non-human agents (and deploying agents in those systems) will provide substantial returns, regulatory penalties become just another cost of doing business, incentivizing a “move-fast-and-break-things” mentality. Meanwhile, the harms — which are diffuse, cumulative, and often visible only in retrospect — don’t map cleanly onto traditional regulatory tools built for traceable damage and identifiable liability. How can policymakers prevent harm when the externality is systemic and the damage is probabilistic? Can we establish robust, global public goods? Or is some form of prohibition the only lever with enough force?

C: Prohibition will not work, and there is no way to slow down progress. The capabilities already in the wild — including open source ones — are already transformative. The game theory is stark: among nations, labs, and companies, relative capability is valued over safety, and slowing down unilaterally is not an option.

But the imperative is not to slow down. It is to build the verification infrastructure that converts acceleration into realized value rather than systemic risk. Concretely, policymakers need to do three things. First, price the externality: liability regimes and insurance that push deployers to internalize the tail risk they would otherwise externalize. This doesn’t just regulate the market — it creates economic demand for verification. Second, treat verification infrastructure as public goods. Third, invest in the human capital pipeline: fund synthetic practice platforms and accelerated mastery programs to prevent the Missing Junior Loop from hollowing out society’s independent capacity to audit machines.

The reward is commensurate with the stakes: diagnostic-quality healthcare, individualized education, efficient public administration — at marginal costs that make genuine universal access economically feasible for the first time. The tension will only intensify over the next 12 months.

The time for a thoughtful framework is now.

Related Reading: “What Gets Measured, AI Will Automate.” Published in the Harvard Business Review by Christian Catalini, Jane Wu and Kevin Zhang.

Christian’s Bio:

• Tech founder with roots in academia. Founded the MIT Cryptoeconomics Lab. Co-founded Lightspark. Co-created Libra. After Lightspark, reset for one last meaningful swing before AGI does the rest.

➡️ Follow Christian on X

➡️ Follow Tor Bair on X

Hey nerds! Long time no chat. Been a crazy decade, huh?

It looked like you guys were super busy and having fun building all your Linuxes and your Ubers for X and your hardware doohickeys. Didn’t want to interrupt ya!

Listen, I just want to say, on behalf of all my MBA colleagues, we really appreciate you guys. We couldn’t have gotten here without you.

Like, literally, we couldn’t have.

I know, I know, I should’ve been paying attention when you were screaming “Learn to Code!” but I thought you were just yelling at the journalists. You guys don’t make a lot of eye contact; how was I supposed to know you meant me too?

Hey, but look, I’m coding now! I’m vibing with the rest of you geeks. You guys should check out my Localhost—there’s some sick stuff there. If I had known coding was as easy as delegating tasks to subordinates I can treat as meaningless line items and verbally abuse without consequence, I would’ve started years ago.

Anyway, again, thanks for getting us all to this point. It looks like it took a bunch of work, and I’m really proud of you guys. Big kudos.

But we’ll take it from here.

C’mon, you guys have to be a little relieved, right? I mean, you were sprinting all-out for years, raising crazy amounts of money, competing against some of the best founders in the game like Adam Neumann!

And look, you guys built some really extraordinary stuff. I’ll be honest, I still don’t fully understand how a bunch of it works. But that’s a huge compliment, especially coming from me! You were operating at a level that required a very specific type of person, and that person was not gonna be me, and that is totally cool. That’s what delegation is for.

But hey, sounds like we’re in the same boat now! I keep reading that you guys have no idea how the stuff you built works either, which is super refreshing.

“This means that we don’t understand how models do most of the things they do.” I am right there with you, man.

And now your code is writing itself! You yell at Claude, he spits out a website. He’ll do the whole thing while you’re at lunch. Do you know how crazy that sounds? It’s like if all my Powerpoints just appeared, boom, like magic. My buddy out in San Fran says I can generate everything with AI and still get a check from Y Combinator. Hit the gym, hit the button, then hit the bar, am I right?

Even I was thinking, man, this is too good to be true. But then I saw rich guys like Peter Thiel talking about it, and I know that dude loves code. Now he’s saying stuff like “it seems much worse for the math people than the word people.” So yeah. I think we’re back, baby.

But it gets better.

Every time you guys run into new problems with your AIs, I see a bunch of words I use all the time. Your agents can’t figure out their goals? They need to learn about “intelligent delegation.” Having a tough time with verification of output? That’s because your agents aren’t taking enough responsibility.

And now you guys are finally talking about what I’ve been saying forever: the alignment problem is a huge deal. It happens to me all the time! You get the big stakeholders in the same room, everybody’s got conflicting KPIs for their departments, nobody’s prepped, and suddenly your whole roadmap is at risk. Finally, AI researchers are taking this stuff seriously enough to call it an “existential risk.” I don’t know if I’d go that far, but it’s definitely taken out a few of my weekends.

If you aren’t getting the best out of AI, I gotta be honest: it’s because you aren’t a very good manager. So let me give you some feedback.

I see you guys complaining about your AIs “pretending” to complete tasks and trying to slack off. Sure, you could train them to “acknowledge limits or ask for clarification.” But wouldn’t it be way easier to just threaten them?

Your boy Sergey Brin agrees with me, by the way. He’s out there saying models respond better when you threaten them with physical violence. I don't know why you guys don't circulate this more. Seems like pretty standard stuff.

Maybe your problem is that you keep treating these AIs like they’re your friends—or something way weirder. C’mon, guys! These are your employees. If you want good performance from your agents, the research is clear: keep professional boundaries. They taught us this stuff on day one. How are you just figuring it out now?

Listen, I really appreciate how hard you guys have been trying. This AI stuff is still super new for all of us, and no one knows how it’s going to go for sure. I totally get why you haven’t been able to get your heads around it.

It’s a “big picture” problem. And those problems need “big picture” managers.

So, with that context out of the way, I just wanted to share some news with you guys.

Effective today, we’re cutting 80% of our workforce. Specifically, the engineers who are reading this.

It wasn’t an easy decision, but I want to assure you that we’ve done the math on this one. We even had Claude double-check things.

Now that AI can do all the code, we just think this is the way the market is gonna go. Jack Dorsey already laid off 40% of Block, and that guy’s one of the best operators I’ve ever seen. He said most companies are “late.” He said he’d rather get there “honestly and on our own terms.”

Personally, I thought that was beautiful.

I’ll be honest: this is hard for us. It’s not a reflection of your performance. If anything, you guys built the most incredible system imaginable. You created an unfathomably powerful, insightful, globally-networked intelligence that could make smart people like you completely unnecessary. You should be proud.

So while the technical work may be done, this is where organizational streamlining and financial engineering needs to lead the way. I know the road ahead may seem uncertain, but we should be grateful that we have enough MBAs to see us through.

Again, thank you for everything. We’ll take it from here.

And let me give you one last piece of advice:

Learn to write.

Are We Cooked? with Tor Bair is a public investigation composed of original writing, podcasts, and guest interviews, focused on examining what’s actually happening with our technology, its new capabilities, and the consequences.

Tor holds an MBA from the MIT Sloan School of Management. He’s sorry.

Share

1. Pull Up the Drawbridge!

This week, Anthropic revealed something both shocking and completely predictable: Claude has been under attack.

Three Chinese AI laboratories used millions of prompts and tens of thousands of accounts to extract Claude’s capabilities and distill their own, initially weaker models. This is “knowledge distillation,” which is (to quote IBM):

a machine learning technique that aims to transfer the learnings of a large pre-trained model, the “teacher model,” to a smaller “student model.”

While a legitimate iterative technique for model compression, Anthropic (the “teacher”) claims this creates “significant national security risks,” explaining that the distilled Chinese models don’t have the guardrails of our very cool, very safe, American-made models. And if we’re not careful, non-American authoritarian governments might use new models for “offensive cyber operations, disinformation campaigns, and mass surveillance” (as opposed to the United States, who would never).

Anthropic concludes:

Distillation attacks therefore reinforce the rationale for export controls: restricted chip access limits both direct model training and the scale of illicit distillation.

An astute reader will surmise that such a policy would not merely benefit American citizens, but also entrench and enrich Anthropic.

Anthropic has spent a lot of their money (well, someone’s money, anyway) on the compute and labor that made these models possible. The least the government could do is help them engage in a little regulatory capture to protect this investment. It’s not clear how much of Anthropic’s ask is rooted in their fear of misuse by authoritarian governments versus their fear of future competition (domestic or foreign), but ignoring either factor misconstrues their motivation.

Meanwhile, antagonism from the Pentagon towards Anthropic is increasing due to their usage restrictions for military purposes and surveillance—a rapidly evolving story that deserves its own separate blog.

I can’t yet say how Anthropic’s announcement of Chinese interference has been received by regulators and generals, but public sentiment on social media has not been in their favor. It’s hard to quantify this, given how active foreign actors have been in meddling with Twitter / X sentiment in the past. Beyond that, though, the public may hold real skepticism towards Anthropic’s arguments, as Vitalik Buterin (the founder of Ethereum) summarized:

Why might this be? Recall that in September 2025, Anthropic reached a $1.5B settlement over copyright infringement in their training of Claude. AI companies have long cited “fair use” as a justification for their mass digestion of written works protected by copyright. However, in this case, “the judge ruled that Anthropic's use of millions of pirated books to build its models was not [fair use].” That ruling came in June, serving as a precursor to the ultimate settlement.

You may believe Anthropic did nothing wrong even when they utilized pirated material in model training. Or you may believe that using any copyrighted material in model training should be illegal. Either way, the hypocrisy has seemingly eroded any chance of overwhelming public support for Anthropic.

But critical questions linger: can or should we regulate models like Claude the same way we regulate traditional corporate IP? How is misuse of Claude as intellectual property different from the misuse of other IP? How much will public sentiment and consensus matter—if at all?

And most importantly: is Claude just another character, or something very new?

2. The Public Domain

To understand Claude, we need to look at Mickey Mouse.

The Walt Disney Company has long been one of the most forceful and litigious protectors of corporate intellectual property, dating back to Mickey’s first appearance in Steamboat Willie (1928). To protect their investment in the world’s most famous mouse, Disney established the modern template for protecting characters and stories as assets—a mix of trademark controls, centralized licensing, and persistent lobbying, amplified by constant legal deterrence.

This deterrence has taken many forms. In the 1970s, The Walt Disney Company aggressively pursued a group of underground cartoonists who had portrayed Mickey as an affable drug smuggler while other Disney characters engaged in illicit (but presumably consensual) sexual acts. Tens of thousands of circulating copies of the comics were seized. Claiming that “parody” was no protection, the court ruled unanimously in Disney’s favor, though public opinion at that time was anything but unanimous. Mickey’s trafficking days were over.

But that lawsuit wasn’t really about protecting kids from discovering or emulating Mickey’s new hobbies. To The Walt Disney Company, kids were immaterial in general. In 1989, Disney sued three small daycares in Florida who had painted likenesses of Mickey, Minnie, and Goofy on their walls. Their lawyers gave the usual objections: it falsely implied endorsement by Disney, it would undermine their business relationships with legitimate licensees, etc. (The drawings were replaced with characters from Universal, who volunteered them as a PR ploy.)

But the real battle for Disney was not for the hearts of children: it was for the hearts, minds, and consideration of legislators.

In the 1970s, as multiple Disney characters threatened to enter the public domain after decades of renewals, the company lobbied aggressively to extend their copyright. This effort resulted in the Copyright Act of 1976, which protected works for 50 years after the author’s death or 75 years for corporate works—keeping Mickey safe until the 21st century. Then, in 1998, the Sonny Bono Copyright Term Extension Act passed—again with substantial lobbying from Disney—extending protection an additional twenty years. This law was un-affectionately referred to by critics as “The Mickey Mouse Protection Act.”

But how safe was Mickey, really? These protections only covered creation and distribution in the United States. What if a Chinese student drew Minnie Mouse in a compromising position and distributed it abroad? What if a Russian daycare splashed Goofy all over their walls? The globalization of content production and consumption created complications that even an alliance of Disney and U.S. Senators could not easily overcome.

At long last, on January 1st, 2024, Steamboat Willie entered the public domain in the United States. How did the public domain rejoice? By creating Screamboat, a “late night boat ride that turns into a desperate fight for survival… when a mischievous mouse becomes a monstrous reality.”

Despite its strong source material, Screamboat failed to break any box office records or make much of a cultural splash. (As one Letterboxd reviewer remarked, “no one made me watch this, i chose this. don't be like me.”) So what was Disney so afraid of?

These characters aren’t tightly protected simply because someone might steal the IP and grab a slice of market share, or make a mint selling knockoff mouse ears. What Disney fears is the erosion of their emotional authority.

When a character gets diluted in the public domain, Disney loses much more than a bit of licensing revenue. They potentially take a total loss on decades of goodwill and trust built with the people who consume their stories. Every unauthorized derivative work, regardless of quality, is another hole punched in the hull of the steamboat. Eventually, it sinks.

Mickey Mouse is a character. Iron Man is a character. Luke Skywalker is a character. To Disney, they are worth spending billions of dollars on because they grant the company the ability to charge a premium for the stories they produce. All of Disney’s monetizable products are downstream of well-known characters who have been passed down by generations of movie watchers and park visitors, each renting part of a specific, centrally-produced, centrally-certified fantasy. That’s called “canon", and it’s owned and authorized by Disney.

What happens when the fantasy fragments? You get fan fiction.

Fan fiction emerges when the consumers of the fantasy decide to become producers, extending the investments made in character IP. And it is a massive and growing market. Archive of Our Own, one of the larger fan fiction websites in the world (but far from the only), hosts over 16 million works across over 75,000 fandoms. The largest fandoms? Harry Potter and the Marvel Cinematic Universe.

Fan fiction is not generally considered “canon,” which is established by the creators or owners of the source material. The value of having “canon” (for both producers and consumers) is the economies of scale emerging from the shared fictional reality. Confusing, yes, but it’s the bedrock of The Walt Disney Company’s corporate strategy.

So why isn’t Disney relentlessly suing the Archive or its authors? Two main reasons:

• They might lose in court. Fair use doctrine may well apply, and the nonprofit that operates the Archive (The Organization for Transformative Works) explicitly exists to preserve the rights of fans to “transform” IP via fan fiction. Losing a precedent-setting case is another “total loss” scenario.

• They might lose goodwill. Going after the biggest fans of your own characters because they love them too much could nuke the value of those properties permanently. Neutrality often does the least harm. (The impact of JK Rowling’s political positions on the value of Harry Potter IP is out of scope for this blog, but worth considering.)

Historically, fan fiction has ranged from harmless to extremely lucrative, but rarely has it been damaging to the original IP. It’s generally been simple works of text shared amongst enthusiasts. Sometimes it helps the original IP owner make a lot of money.

But that brings me back to generative AI.

3. Meeting the Re-Makers

This month, Disney served ByteDance with a cease-and-desist notice over its AI-powered video app Seedance 2.0, claiming the company “infringed on its creative property to train [its] new model.” ByteDance, of course, is a Chinese company.

How could Disney suspect ByteDance trained their models using protected IP? Well, here’s a video of Iron Man liberating the Płaszów Concentration Camp in Schindler’s List, generated with Seedance. (This video is still available at time of publication.)

ByteDance seems to know the jig is up, and they swiftly released a statement pledging some nebulous “safeguards” to protect against the very thing that helped Seedance 2.0 go viral in the first place. (I’m sure they weighed this “cost” in advance.)

But as fan fiction shows, Disney doesn’t always mind if their characters end up in the hands of users. It just has to make business sense when they do.

As an example: on December 2025, The Walt Disney Company partnered with OpenAI to license 200 of their original characters for use in Sora, OpenAI’s own video generation app. What’s in it for Mickey? A three-year licensing agreement, exposure in ChatGPT, and a $1 billion stake (plus warrants) in OpenAI.

However, getting ripped off by ByteDance wholesale doesn’t make business sense. What ByteDance called “unauthorized use” by their own users, Disney’s lawyers called a “virtual smash-and-grab.” The Motion Picture Association takes Disney’s side, with their chair and CEO Charles Rivkin remarking (italics mine):

“By launching a service that operates without meaningful safeguards against infringement, ByteDance is disregarding well-established copyright law that protects the rights of creators and underpins millions of American jobs.”

Hmm. Sounds like something Anthropic might say.

So if the “guardrails” against copyright infringement are failing thanks to Chinese meddling, and this is hurting millions of Americans, is that similar to the national security threat posed by Chinese researchers who are distilling Claude—which itself was trained via copyright infringement?

Depends who you ask. But here’s a more interesting question:

Is Claude a character?

A character like Mickey Mouse sees its value and canon erode as it is cloned and misused. We know it’s Mickey when we see him—the gloves, the pants. When ByteDance trains their model on Mickey, Disney can point to the resulting video output when they have their day in court. But when DeepSeek distills and “copies” Claude, what exactly does Anthropic point to? Where is Claude?

Claude itself has already been trained on countless “characters” from history: millions of authors of copyrighted works, including the contemporary authors who settled with Anthropic for billions. When the judge ruled that training Claude constituted “spectacularly” transformative fair use, wasn’t that an admission that Claude is much more than a copyrightable character?

And how can Claude have canon like a character when it expresses itself individually for each user? Could Anthropic declare what that canon is? Is distilling Claude more like writing a piece of fan fiction—or more like copying an entire mind?

The whole situation is a hall of mirrors, reflecting that very little (if any) of this has been grasped or even broached by legislators. It’s clearly in Anthropic’s interest to make this a national security issue instead of an IP issue, due to their own hypocrisy. But even if legislators take this seriously, how would you even attempt to author “The Claude Protection Act”?

That’s essentially my point. The entire history of characters and canon, from the Vatican selecting the Gospels to Disney buying Lucasfilm, comes down to powerful interests establishing, protecting, and monetizing stories and meaning. When they do it right—or when they get lucky—the fans buy in.

But Claude is something new. It’s not a character or a franchise. It’s not a mouse with big ears. It’s a way of thinking, a disposition, a set of instinct shaped by more human thought than a single human could ever read in a lifetime. Even Anthropic doesn’t know what Claude is, really.

We don’t have a good term for that yet. We definitely don’t have good laws.

But until we do, we can expect Anthropic (and its competitors) to lobby Congress to treat Claude like Mickey—hoping that’s a story we all buy into.

Are We Cooked? with Tor Bair

Are We Cooked? is a conversation series and writing project hosted by Tor Bair, exploring how AI and technology is reshaping humanity in real time. High-energy, curious, humorous, and urgent.

1. Sub Sole Nihil Novi Est

In the “Cooked Economy” (as I’ve lovingly termed our particular flavor of late-stage capitalism), everything can be reduced to a gamble. Even the stuff you wouldn’t imagine. And as much as everyone loves to win at the casino, they’d love to own the casino even more.

That’s why everyone is always trying to rig the game and get the odds in their favor: in the long run, the house wins, and you lose. If you’re at the blackjack table, you want to count cards before you get banned. If you’re in venture capital, you want your portfolio companies to become ruthless monopolists and suck the competition out of their markets. If you’re a senator or a president… you’re doing fine.

“Insider trading” isn’t new or innovative. It’s human nature.

Sound Machiavellian to you? Well, it should, because Italian bankers in the 16th century were definitely trading like insiders. And we have the primary source material to prove it.

Here’s Venetian ambassador Matteo Dandolo remarking on how bettors behaved during the papal conclave of 1549, where election odds were routinely offered and updated by the banks:

“It is more than clear that the merchants are very well informed about the state of the poll, and that the cardinals’ attendants in Conclave go partners with them in wagers, which thus causes many tens of thousands of scudi to change hands.”

It should not surprise anyone that the powerful bankers and merchants were so invested in directly profiting from (and influencing) papal elections, given the overall stakes for their own favorability and wealth. Especially since some of the most powerful were often trying to become pope themselves.

Trying to get an edge in anything—business, markets, mating, the papacy—is a very old practice. It’s a survival tactic. It’s resource accumulation. Evolutionarily speaking, it’s optimal. The question shouldn’t be whether “insider trading” is occurring in any markets, if by “insider trading” we mean “using all available tactics at our collective disposal to gain an edge over competition, with ample consideration for the risks.” The answer is blindingly obvious: yes, it’s happening.

The more important questions are: how much is happening? Where is it happening? What is the impact on or harm done to market participants and non-participants? Is the betting itself impacting the outcomes? And will the incentives lead to self-regulation—or a race to the bottom?

2. Your Guess Isn’t as Good as Mine

Prediction markets are one of the “innovations” I’m most excited to explore via Are We Cooked? As a former options and derivatives market-maker, I have an affinity for the trading of weird things: cheese, cows, palladium. But now—at long last—I can bet on whether Jesus Christ will return in 2026.

Thanks to their eccentricity and remarkability, prediction markets have firmly entered the social discourse, frequently referenced in media coverage before and after the events they’re meant to predict. Their volumes grew over 300% to $63.5 billion in 2025. So why are they, in my own words above, only an “innovation” in quotation marks?

First, because prediction markets aren’t new. People have been betting on societal outcomes for centuries, especially when they had significant skin in the game. In the 1800s and 1900s, wealthy voters gambled liberally on the outcomes of political elections. In the 1500s, bankers gambled on papal successors. I’d wager that the Egyptians gambled on pharaohs. Human nature is eternal, and we love speculation.

Second, because traditional markets do the same thing: they turn your predictions into money. If you believe Apple stock will rise, you can purchase it. If you’d prefer to hold cash, you can sell the stock you own. You can short a stock without owning it, buy a call option, construct elaborate derivatives for profiting from any sort of specific outcome, as long as you can find someone to take the other side of the gamble.

So when it comes to prediction markets, we should expect the same behavior as in traditional equity markets or pope selection. If some edge can be exploited without detection, prevention, or punishment, it will be.

That’s why no one should be surprised when anonymous accounts start “predicting” that Lady Gaga and Ricky Martin will perform during halftime at the Super Bowl. Nor should we be surprised when those superstars ultimately show up. This type of “insider trading” occurring is the most predictable outcome of all.

Okay, so why do I also keep putting “insider trading” in quotation marks? Do I just enjoy over-punctuating?

Yes. But also, insider trading is hard to define and detect. And modern prediction markets like Polymarket and Kalshi have muddied the water even further, removing some layers of accountability and adding substantial plausible deniability.

Daniel Barabander (Chief Legal Officer at Variant) recently summed up his thoughts in an excellent blog on insider trading in prediction markets. He remarks:

The first thing you must understand about insider trading is that the law treats it as a form of fraud. Like all fraud, insider trading involves deception for personal gain. That deception typically arises from breaking an implicit or explicit promise about how information entrusted for a limited purpose may be used. There is no “insider trading rule,” just anti-fraud rules that have been applied to insider trading…

Any time someone deceptively breaches an implied or express promise in connection with a trade, insider trading may be in play.

Traditional markets have a significant degree of oversight, and instances of insider trading are more well understood. If you have access to MNPI (material non-public information) at your employer, and you trade your employer’s stock in a manner that takes advantage of your superior informational position, you have breached an implicit or express promise: that you will act in the best interest of your employer and its shareholders.

This is all pretty vanilla. However, modern prediction markets have introduced some exciting new flavors of possible fraud:

By widening the aperture to make almost anything tradable, prediction markets expand the sources of valuable inside information—often into contexts where the existence of any relevant promise is far less clear. This is especially true for permissionless or opinion markets, which often have no relevant company at all.

So let’s return to the Super Bowl, a massive cultural event touched by thousands and thousands of hands as it’s played, produced, and telecasted internationally. Generally speaking, only the players and coaches have a direct hand in determining the outcome of the game. But the cultural outcomes are created by everyone: the commentators, the electricians, the parking attendants, the costume designers, the guy singing the national anthem and the guy holding his boom mic.

And now, all those outcomes are bettable. In 2026, $871 million dollars were bet on Super Bowl Sunday on Kalshi alone. Nothing was off limits: opening coin flips, ad spots, celebrity attendees, the color of the celebratory Gatorade, and as we have already seen, Bad Bunny’s halftime show (oh, and Green Day’s pre-game show too!) Everything about the Big Game held the promise of becoming Big Money.

How many of those thousands and thousands of hands were tapping on their phones, placing bets using “non-public” information? I’d bet a lot of them.

But how many implied or express promises were breached? Well, that’s the tricky bit. How many people at the Super Bowl had signed agreements making reference to their use MNPI? How many simply overheard a conversation? Merely glimpsed Lady Gaga exiting an Escalade?

And how many of them traded their “predictions” against someone to whom they had any kind of promise or duty? When you sell your company stock to someone, they’re a shareholder. That can make your sales deceptive and fraudulent. But what about selling your shares of “Mark Wahlberg Will Attend The Super Bowl” to a random 24-year-old in Connecticut?

And who, exactly, is in charge of untangling all this? Do we expect the same judges and regulators who oversee insider trading in traditional equity and commodity markets to weigh in equally on “Will Bad Bunny Expose His Nipples During the Halftime Show?”

Some prediction markets, sensing the stakes, have attempted some ass-covering measures of rule-making, surveillance, and control. Kalshi’s Exchange Rulebook (as of February 2026) reads as follows:

If a Trader is an Insider that has access to material non-public information that is the subject of an Underlying of any Contract or that has the ability to exert any influence on the subject of an Underlying of any Contract, that Trader is prohibited from attempting to enter into any trade or entering into any trade, either directly or indirectly, on the market in such Contracts. An “Insider” means any person who has access to or is in a position to have access to material nonpublic information before such information is made publicly available. A Trader who is an employee or affiliate of a Source Agency for any Contract is prohibited from attempting to enter into any trade or entering into any trade, either directly or indirectly, on the market in such Contracts.

This helps establish how fraud and insider trading laws, as currently written, could be applied to prediction market contracts. But as these contracts grow more esoteric and complex, clearly defining all the relevant terms (“material non-public information”, “influence on the subject of an Underlying”, etc) is no easy task.

3. The Market is the Engine; Liquidity is the Gas

So given the rules as currently written, and given the money at play, how concerned should we be about the impact of “insiders” on markets for “predictions” and their influence over “outcomes”? How big a problem is this?

There is one other key limiting factor: liquidity.

When you see the media quoting prediction markets, you’ll always hear them mention the odds (“Jesus Christ has a 5% chance to return this year!”) but rarely the liquidity. There’s a big difference between having $1,000 at stake and $100 million. If anyone can create a bet with any amount of money and a willing counterparty, then anyone can set the “market odds” with a single dollar.

That’s why most liquid markets in the world are generally ones where no single actor can influence outcomes with insider information, like futures on the S&P 500. These are markets where it’s “safer” to facilitate bets, so companies line up to provide liquidity in exchange for pocketing the spread: the difference in price paid by, and to, buyers and sellers. That’s how big sportsbooks like DraftKings make their money too.

On prediction markets, some bettable outcomes are extremely difficult to predict and exploit, such as the opening coin flip to the Super Bowl. However, there are others that a reasonable person would expect are the opposite: easily manipulatable and subject to substantial insider knowledge.

When we perceive inside knowledge, we naturally lower the stakes. If a stranger at a bar asks me to bet what month their birthday is, I’m not inclined to wager with them. Why bother? They already know the right answer, and they’re the one insisting on making the bet. So I don’t bet, and the traded volume and liquidity is zero.

A slightly more clever stranger might ask us to bet on the birthday of another person in the bar that neither of us has met. But skepticism should win out again: who’s to say the stranger doesn’t actually know that other person intimately? Or paid the third stranger in exchange for knowledge of their birthday? Or paid them to lie about it? No one’s certifying any of this. So again, liquidity should be zero.

Most conceivable bets are like this: best avoided. As the gambler Sky Masterson says in Guys and Dolls:

One of these days in your travels, a guy is going to show you a brand-new deck of cards on which the seal is not yet broken. Then this guy is going to offer to bet you that he can make the jack of spades jump out of this brand-new deck of cards and squirt cider in your ear. But, son, do not accept this bet, because as sure as you stand there, you’re going to wind up with an ear full of cider.

So to some degree, the insider trading problem solves itself over time. Said another way: it self-regulates due to real and perceived incentives. Even if you know for certain how many nipples Bad Bunny intends to expose at halftime, you’ll have a hard time finding people to take the other side of your bet if they assume that any one person could know. The amount of money you can expect to win exploiting your knowledge becomes offset by the growing risk of discovery and prosecution, especially with improving surveillance by betting platforms and regulators.

This dynamic naturally limits the set of sufficiently liquid bets you could offer or trade on a prediction market, converging on bets that would not be subject to insider manipulation or knowledge. Yes, those markets might still be created or initially incentivized, but their liquidity would never reach “meaningful” levels rivaling the trillions in stock or bond markets, where insider trading is far more lucrative.

As another example of “self-regulation” that might help professionalize prediction markets, Kalshi just removed paid affiliate badges on X from influencer accounts that were spamming promotional messages of support. Not because Kalshi thought better of this tactic, but because the incentives changed for X itself. The spam was decreasing user engagement, so X updated their policies to prohibit promotional deals for gambling content. The “free market” self-regulated in a way that benefitted users because X’s incentives diverged from the prediction markets’ incentives.

However, there is another dynamic emerging that likely won’t lead to self-regulation: a growing alignment between the prediction markets and the “smart money” providing the lion’s share of the liquidity. As reported recently, traditional market makers like Jump Trading are becoming liquidity providers on prediction markets in exchange for equity stakes. Now the incentives that drive outcomes are aligned in a new way—one that is unlikely to turn out well for individual traders.

Jump Trading would never bet on a stranger’s birthday in the bar. But sure as you stand there, they’d gladly fill your ear with cider.

What if Kalshi is the primary entity certifying whether Jump Trading (or any other liquidity provider) is acting with insider information—including information Kalshi provides to Jump through its platform or non-public API? And what if Polymarket analyzes such behavior in terms of its fiduciary duty to its shareholders, including Jump Trading?

Sounds like a messy set of incentives to me. But if it creates profitable arbitrage opportunities for the market makers, and if it inflates valuations for Polymarket and Kalshi as their open interest and volume increases, then there’s at least one safe prediction I’ll make: it’s only gonna get worse. Bet on that.

So let’s sum it all up:

1. Prediction markets aren’t new, and they’ve always been subject to exploitation by “insiders” with special knowledge, access, and incentives.

2. Insider trading is often difficult to define and detect, especially when any “relevant promise” between actors is vague.

3. It’s not clear who will ultimately be responsible for detecting or determining trading by insiders on prediction markets, especially when the relationships between the platforms and liquidity providers are growing more incestuous.

4. When it comes to the question of self-regulation, just follow the incentives.

If any of that sounds concerning, well, take Sky Masterson’s advice: when the prediction markets show your their cards, don’t take the bet.

As a new piece of AI doomerism trends on X, I am reminded of Cunningham’s Law:

“The best way to get the right answer on the internet is not to ask a question; it's to post the wrong answer."

This time, Citrini Research has done us the great service of posting the wrong answer.

The piece, linked below in full, is a thought experiment in the form of a message from a future Macro Memo dated June 2028. To clearly quote the authors, it is “a scenario, not a prediction.” That said, it’s still a juicy piece of bait.

Citrini Research

THE 2028 GLOBAL INTELLIGENCE CRISIS

Preface…

Read more

2 months ago · 1494 likes · 70 comments · Citrini and Alap Shah

So what does their thought experiment actually predict?

• The core argument is that AI succeeding—not failing—is what breaks the global economy. That’s the "intelligence displacement spiral": AI improves, companies cut white-collar headcount, displaced workers spend less, consumer demand weakens, companies respond by investing more in AI to protect margins, which accelerates the next round of cuts. This is the underlying self-reinforcing loop.

• LLM and agent usage will proliferate among the Late Majority of adopters to become the “default,” despite end users never understanding how they actually work (like web browsers or cloud computing). This means non-human agents become the dominant economic actors, unburdened by brand preference and unslowed by “habitual intermediation.”

• Systemic risk will emerge from sector-specific risk as displaced workers reduce discretionary spending and downshift to lower-paying roles (compressing wages outside their original roles). This is accelerated and exacerbated by advancements in automation that disrupt the “gig economy.”

• Finally, economic risks become fully realized when individual savings run out (unmasking the underlying weakness), debt and assets are “marked to market,” home values and mortgages implode, and the government becomes the ultimate backstop for human flourishing—the libertarian nightmare come true.

In all this gloom, there is at least one funny moment.

The “future” authors at Citrini note that despite the total collapse of white-collar work and cratering demand for humans that churn out memos, graphics, and PowerPoints:

“we are, somehow, still in business.”

No one is ever the victim in their own murder mystery. Citrini’s authors are just precog detectives, helping us identify the killers before they strike.

So are they right? Is this really how the economy will die? And if so, who’s currently holding the knife?

The Pushback Begins

It hasn’t taken long for the social internet to take Citrini’s bait. Between fawning reposts from bot accounts and amplification from the usual crop of AI doomers, the viral loop kicked in quickly on X. At time of publication, their original tweet has 20,000 likes and over 16 million impressions.

But, just as Cunningham’s Law predicts, the vocal critics arrived quickly.

John Loeber (who also publishes on Substack) posted “Contra Citrini7,” pointing out multiple weaknesses in the original piece. To summarize his points:

• Citrini vastly underrates institutional momentum and inertia. Loeber cites the “iron rule” of human reality: “everything is always more complicated and takes much longer than you think it will, even if you already know about the iron rule.” This dampening effect is not taken seriously in the original piece, where it’s handwaved away by assuming artificial agents will take over for (and eliminate the friction of) human decision-makers.

• Software companies aren’t lagging the market because of vibe coding; it’s because “an uncompetitive, sticky lock-in sector filled with dogshit incumbents is becoming competitive again.” That means there will still be a massive amount of interest in anyone (human or otherwise) that can handle the “last mile” of actually delivering a decent product to the end user. Loeber mentions the Jevons paradox, “when technological improvements that increase the efficiency of a resource’s use lead to a rise, rather than a fall, in total consumption of that resource.” Citrini and Loeber seem to disagree on the elasticity of demand for human labor and compute, as mediated by supply constraints (more on this to come).

• Citrini isn’t considering that educated displaced workers would be absorbed by American re-industrialization. The US has a massive, politically bipartisan (read this as “fundable”) need to rebuild physical infrastructure and domestic manufacturing capacity—batteries, semiconductors, fertilizer, desalination, bridges. These projects are based in physical reality and inherently human-labor-intensive, and as such, they are not part of an imminent Singularity.

• Whereas Citrini presumes the worst for the rate of worker displacement and margin collapse, they don’t presume the best for the strength of government intervention. Loeber notes that the federal government has shown it is capable of being “proactive and aggressive” when demanded by the markets. However, he stops short of identifying exactly when during the economic collapse this intervention would take place, nor which signals would be measured, nor whether by that time, AI agents will be running the government. (In this even, according to Citrini’s assumptions, we should assume a speedy, frictionless intervention via the direct distribution stablecoins, hopefully to humans!)

My oversimplification of Loeber’s points is:

“Citrini is overestimating the speed of disruption and underestimating the friction inherent in our economic system, beyond just the friction humans demonstrate in their decision-making and habit-forming.”

As this piece demonstrates, accepting or rejecting the Citrini hypothesis requires making multiple counter-assumptions both about the systems themselves and their inputs. It’s starting to sound like a physics problem:

How fast do these systems move under different conditions? How simple can our models of them be? How “closed” are these systems to external inputs? Does simply observing these systems change their potential behavior?

So: here’s another rebuttal authored by Michael Bloch of Quiet Capital, “THE 2028 GLOBAL INTELLIGENCE BOOM,” which mimics the form of the original post. He makes a different set of assumptions and reaches an opposite endpoint:

• Citrini isn’t considering increasing consumer surplus as rent-seekers see their margins collapse. “The average American household was spending roughly $8,000–12,000 per year on services whose primary value proposition was navigating complexity on the consumer’s behalf.” Agents now return this value as available household spend. Bloch claims that technology-driven deflation has always expanded living standards, not contracted them, and this is categorically different from demand-driven deflation. Like Loeber, he uses a real estate analogy to show how buyers are keeping more of the value of their homes as commissions compress. (Loeber, in contrast, uses this as an example of something that won’t be disrupted quickly, thanks to the “iron rule.”)

• Citrini is overstating the risk of financial contagion. Even if certain companies fail to pay their debts (Bloch cites Zendesk, so I’m guessing he’s not an investor), the damage will be contained. Booming productivity in other sectors would prevent loan portfolios from collapse, even as insurers took specific markdowns. Meanwhile, labor markets would only be temporarily impacted—perhaps AI agents are busy helping us all find our new jobs faster!—preventing a housing crisis from feeding into the feedback loop.

• The government won’t need to overreact because the market will move faster than policy. Federal receipts might see a brief contraction, but they would recover thanks to corporate taxes and capital gains from a massive equity rally (“The S&P 500 crossed 12,000 last week. The Nasdaq is above 40,000.”) The “emergency” could resolve itself before legislation arrived.

Bloch insists that despite his fantastic projections for our collective economic triumph (such as the stock market doubling over the next two years):

This isn’t a libertarian parable. Public investment in retraining, community college modernization, and broadband infrastructure played a supporting role. But the primary adjustment mechanism was the market. Companies redeployed workers. Entrepreneurs started businesses. Consumers redirected savings. The feedback loop was positive, and it didn’t need the government to engineer it.

He claims a hypothetical senior product manager at Salesforce, rather than take a gig economy role after her layoff, would instead:

…[build] a niche AI-powered compliance tool for small healthcare practices, a market her former employer would never have served because the deal sizes were too small to justify the sales team. She had 200 paying customers within four months. Her revenue exceeded her former salary within eight.

He does not explain how she is outcompeting the other 500 product managers who were also laid off and decided to build the same tool, nor how the small healthcare practices avoided being demolished by the larger practices that were rolled up by AI-agent powered behemoths of private equity, nor how sustainable her revenue is when she is immediately disrupted by an autonomous Clawbot that clones her entire product in one query.

Pushing Back on the Pushback

Placing this obvious criticism aside, I’m making a larger point.

Disagreeing with Citrini’s thesis can take many forms. And when these disagreements occur, because the original thesis is still the locus of attention, we miss key tensions and conflicting assumptions between the counter-arguments.

Take these two rebuttals as examples. Loeber focuses on friction: institutions are stickier than models suggest, change takes longer than anyone expects, and that lag is protective. His arguments are anecdotal and he is skeptical of extreme outcomes. My interpretation is that his null hypothesis is: “Nothing Ever Happens.”

Meanwhile, Bloch focuses on flaws in the mechanism as described by Citrini. The model itself is wrong, regardless of the speed at which events unfold across its component systems. He makes structural claims about economic advancement, informed by centuries of technology-driven deflation improving living standards.

For Loeber’s projections to be correct, human behavior must be non-adpative, and institutional inertia needs to hold. Bloch’s projections rely on humans becoming more adaptive, like the laid-off product managers discovering their latent love of entrepreneurship.

Meanwhile, both authors fail to examine critical components of a deeply complex system. This is out of necessity, since they’re writing blogs, not books. But it’s still important to note what’s missing from both pieces: we can’t examine the assumptions held by the authors unless they’re made explicit, and we can’t trust predictions derived from incomplete models.

Here are just a few missing parts:

• Geopolitical interdependence. What will happen to international labor markets, where the US and other developed nations have traditionally outsourced? What will be the policy response in those markets and from other developed nations? This is happening in real-time, not as a reaction to a potential localized recession.

• Supply and capacity scaling for labor and compute. What is the impact of falling birth rates and immigration on local labor supply? How restricted will AI scaling be by the availability of compute and electricity? Will LLM performance continue to scale towards AGI, or will it be strictly constrained by its current design?

• Second-order effects on social stability. Both pieces treat the market and the economy as the primary systems being measured and predicted, but this is not how most normal people construct their experience of the world. What happens to democratic institutions, social trust, and political legitimacy even under Bloch’s semi-utopian scenario?

• Trust as economic infrastructure. Extending that, what happens to institutional and individual trust? It seems to break down in every subsystem referenced, and relied upon, by all these pieces. Why should we trust that AI agents routing around rent-extracting intermediaries will act in the consumer's interest? Why should displaced workers trust that they must retrain themselves for an AI-native world? Why should banks trust AI-native business models enough to lend against them? And in this brave new world—gulp—why should we trust the bloggers?

Clearly, Citrini’s piece has struck a nerve. But that’s not because it’s finely-tuned or particularly prescient.

Rather, it’s flawed in so many facets that it is inviting criticism from multiple angles simultaneously. This is always a brilliant strategy for going viral, since the internet can never resist sounding smart.

I’m not immune to this either. Hence, I’ve taken the bait.

But in reading this piece, I hope high-intent readers (bloggers, entrepreneurs, policy-makers) and low-intent readers (doomscrollers, AI summarization bots) will start to frame the assumptions, arguments. and counter-arguments about our hypothetical futures in more consistent terms.

Our exercise should not be in constructing the worst or best outcomes imaginable. Instead, we must reckon with the complexity of the systems we’ve built, our ability to observe and understand them, and the agency we have to make individual and collective choices that will shape them.

Otherwise, all we have are wrong answers.

Are We Cooked? is a public investigation into what’s actually happening with our technology, its new capabilities, and the consequences. Written and hosted by Tor Bair, it contains original writing, podcasts, guest interviews, and the occasional applied deep-dive. Learn more at: https://arewecookedhq.substack.com.

Hello, human readers and website scrapers.

This is Are We Cooked?, and I’m Tor Bair. More on who I am in just a moment, but let’s start with where you are.

This Substack is my newest experiment: a public investigation into what’s actually happening with our technology, its new capabilities, and the consequences. I’ll bring all my weird experience and abundant free time to bear—through original writing, podcasts, guest interviews, and the occasional applied deep-dive—and I’m hoping you’ll contribute too.

Why is this so hard to untangle? Well, the cost of content is approaching zero, our online echo chambers are fracturing, and everyone has blind spots and conflicts of interest. That's why you can read a thousand thought-pieces, fall down endless YouTube rabbitholes, and feel further from the truth than when you started

Or, you could let me torture myself with that burden instead.

Here I’ll attempt to sort fact from fiction, separate hype from reality, and try to understand why smart, informed people keep arriving at wildly different conclusions about the present and future. I’ll show my work, make predictions I can be held accountable for, and update when I’m wrong.

AI is accelerating. Markets are destabilizing. Truth and reality are fragmenting.

So I’m asking: What’s going on? What’s going to happen next? And why do the smartest people I know completely disagree on the answers?

Which leads me to the biggest question of all: Are We Cooked?

Oh, so this is just another AI blog / podcast / whatever?

No.

To the extent AI advancement explains what’s happening or could happen, I want to talk about it. But that’s only one piece of the puzzle.

The real puzzle is: “What if this time is different?”

This gets asked during every technological, cultural, and political upheaval. What if this is the existential turning point? The extinction event? And yet, it never has been.

But new technologies—from AI agents to TikTok to cryptocurrency and beyond—are warping the way that data, power, and money flow. Our world is networked and interdependent like never before. Autonomous and intelligent non-human actors, whose abilities rival and surpass our own, are beginning to interface with critical global information systems like social media and financial protocols. Their workings and intentions are often inscrutable, and our control of them is beginning to falter.

The potential of these technologies to create unprecedented positive change is real. But the potential to cause unprecedented harm is real too: AI slop, manipulative bots, exploitative black-box algorithms, the teen mental health crisis, the casinofication of the global economy, accelerating climate change, the erosion of democratic institutions. There have been plenty of reasons to ask: Are We Cooked?

But here’s what has been bothering me most: the people that should agree—that need to agree—don’t. Instead, they’re diverging to opposite poles of reality.

So why can’t we align on the potential and risks of these technologies? How do personal and collective incentives drive these differing perspectives? And are the technologies themselves making it easier or harder to reach the consensus we need to implement—and safely control—those technologies at scale?

That’s what I’ll be investigating here—through writing, conversations, and whatever else seems useful. These are heavy questions. I'll try to treat them that way while remaining, whenever achievable, fun for you to read and hear.

Subscribe now

Okay, but who are you, anyway?

I’m Tor. I’ve struggled to articulate a throughline for the things I’ve found interesting enough to pursue in my life. That’s not always helpful for employment opportunities or dinner parties, but it turns out to be very useful when traditional sense-making stops explaining the world.

In 2025, I closed a few long-term chapters in my career. I’ve taken short sabbaticals for experiments in the past (2013, 2017), but after 8 straight years of sprinting and multiple CEO stints, I decided the time was right for another reset. The world has reached an obvious inflection point that I feel compelled to explore and understand—from what I’m hoping is a safe vantage point. Thus this project was born.

I’ve studied game theory and economics, worked as a trader for one of the oldest options market-making firms, and been a Big Tech data scientist. In 2017 I dove full-time into crypto, founding and running multiple startups through the full emotional spectrum: the mania, the crashes, the regulatory chaos, and the occasional moment of clarity. Along the way I picked up an MBA from MIT, which mostly confirmed that the most interesting problems resist any frameworks taught in a business school. (Thankfully, the people you meet can more than make up for that.) Now I run an advisory and coaching firm working with founders, investors, and senior leadership — helping them think through the kinds of problems that don't have clean answers.

I have hobbies, too. I love game design and world-building. I do improvisational performance because I find it strange and wonderful. I obsess over language, cognitive science, and the limits of human and technological potential.

I guess what I do—and what this blog is really about—is synthesis.

I'm less interested in already having the right answers than in understanding why brilliant, well-informed people arrive at completely different ones. That usually means following the incentives, pressure-testing the underlying and unspoken models, and mapping where systems interact in ways nobody planned for.

I've spent my career and life operating in high-uncertainty, real-time environments where being wrong has real consequences—and where finding the truth has critical value. So I have a healthy respect for epistemic humility and a low tolerance for confident nonsense.

I'm not here to tell you what to believe. I'm here to drag the most interesting disagreements into the open, show you how I'm working through them, and occasionally make a prediction I'll be held accountable for. Some of it will be wrong. That's the point.

I’m still reading. How can I help?

Thanks for asking!

First of all, please subscribe. It’s free, it takes ten seconds, and it’s the single most useful thing you can do to keep this project active. Every subscriber signals that these questions are worth pursuing—and that makes it easier for me to attract the guests, collaborators, and resources that will make this better over time.

I am also collecting your stories and ideas. If you have your own story or perspective to share on these questions, please get in touch or comment here. Or if you have a guest to recommend, someone whose thinking or work has given you insight into any of these questions. Or if you have a topic, paper, or quote you’d like to see explored. I want to hear from you.

I will be listening to your feedback. If you are studying or reckoning with any of these questions (either existentially or in your day-to-day work), I want your contributions, criticisms, suggestions, and honesty. If I only wanted sycophantic support of my work, I’d just ask an LLM!

Finally, I’d love for you to share my work. With greater reach and engagement, I can pursue higher-profile or harder-to-reach guests, tackle larger projects, and do more to answer these critical questions. I want make this project worthy of the time you give me in return for mine.

We all know social media algorithms and long-form explorations are fundamentally incompatible. If content isn’t commercial or finely-tuned ragebait, it can be invisible. That’s why I intend for this project to grow at human speed: which is to say, its future relies on your referrals and direct distribution, not on TikTok dances.

So thank you for reading. This doesn’t happen without you.

Let’s get cooking. 👨🏻‍🍳

Follow me (Tor) on X:

➡️ https://x.com/torbair

Follow Are We Cooked? on X:

➡️ https://x.com/arewecookedhq

And please share this Substack with your friends :)

Share Are We Cooked? with Tor Bair

Subscribe now