There's a conversation happening about what's coming for labor. It's not happening in recruiting. It's not happening in your LinkedIn feed between the "AI won't replace you" crowd and the people posting screenshots of their latest automation workflow. It's happening in economics departments, in independent newsletters, on Substack, in academic papers, and in a worker cooperative that formed two months ago in a GitHub repository.

Recruiting has its own mythology about labor. The pendulum. The cycle. Buyer's market, seller's market, back and forth forever, and all we need to do is wait for the swing. That story has been useful for a long time. It told us that whatever's happening now is temporary. That the market corrects. That if you're good enough and patient enough, the jobs come back.

Here's the problem: the pendulum is bolted to a wall, and someone is taking the wall down.

This edition is different. I'm not writing about recruiting. I'm pointing you toward the people I think are doing the sharpest work on what's actually happening to labor, structurally, right now. Some of them you've never heard of. None of them are recruiters. That's the point.

Where to find him: futureforwarded.substack.com

Dodson is publishing near-daily AI Labor Reports that are, as far as I can tell, the most rigorous running account of what AI is actually doing to the job market. Not what it might do. Not what Sam Altman says it will do. What it is doing, tracked through WARN filings, earnings calls, BLS data, and the gap between what executives say on investor calls and what they say to the people they're firing.

His most important contribution is a concept he's been developing called "Ghost GDP." The short version: the economy is growing. Corporate profits are strong. And the same quarter that produced those numbers also produced the largest cluster of AI-attributed layoff announcements in corporate history. GDP growth driven by AI infrastructure investment and productivity gains that flow to shareholders does not automatically reach the worker whose coordination role just became unnecessary. The economy is growing and hollowing out at the same time, and our measurement tools weren't built to see both things at once.

If you read one thing from this list, read his May 1st report. It's the clearest single document I've found on why the standard reassurance ("but the economy is growing!") is not the comfort people think it is.

Where to find him: disconnect.blog / Tech Won't Save Us podcast

Marx is a Canadian tech critic, and his position on AI and labor is the most counterintuitive and (I think) the most important one on this list. He does not believe AI is going to wipe out half of all jobs. He's been saying that since before it was contrarian, back when the automation panic was peaking.

His argument is that the real function of the AI displacement narrative is not prediction. It's leverage. The threat of mass automation is a tool companies use to deskill professions, push down wages, and consolidate power over workers. The same playbook ran through the gig economy. The same playbook ran through the self-driving car hype. The technology was never as powerful as the executives claimed. But the labor concessions extracted while everyone was scared of robots? Those were real, and they stuck.

This doesn't mean nobody is losing their job. People are. But Marx's point is that if you're only watching the layoff numbers, you're missing the larger structural damage: the slow conversion of careers into gigs, of expertise into "content," of professional autonomy into algorithmic management. The thing that kills you isn't the robot. It's the boss who uses the robot as an excuse.

For recruiting people specifically, this should land hard. Think about what happened to recruiting itself. The profession didn't get replaced by AI. It got deskilled, squeezed, and platform-ified while everyone was debating whether chatbots could write job descriptions.

Where to find him: aaronbenanav.com

Benanav is an economic historian, and his book drops a bomb on the entire automation debate. His argument: the long-term decline in labor demand is not caused by technology. It's caused by global overcapacity and deindustrialization. The robots are a convenient story. The actual structural problem is that capitalism has been producing chronic underemployment for decades, and automation discourse lets everyone avoid talking about why.

This matters because it reframes the entire conversation. If the problem is technology, then the solution is retraining and adaptation. If the problem is structural overcapacity in the global economy, then retraining is a band-aid on a hemorrhage. Benanav isn't anti-technology. He's arguing that we're diagnosing the wrong disease, and the wrong diagnosis produces the wrong treatment.

His newer work focuses on what he calls the limits of "one-dimensional optimization," the tendency of markets, firms, and algorithms to reduce everything to a single metric. Efficiency. Output. Productivity. The things that get lost in that reduction are exactly the things that make work worth doing.

If you're a recruiter who has ever felt the dissonance between "we need to hire the best people" and "we need to reduce cost-per-hire by 30%," Benanav is articulating what you're feeling at a structural level.

Where to find him: ethicore.substack.com

McDonogh's Substack is focused on AI governance and accountability in marketing, which sounds niche until you read his piece "AI Made Us Do It. Sure It Did." He walks through the spring 2026 layoff wave (Meta cutting 8,000; Microsoft offering buyouts to 8,750; Oracle eliminating 30,000; Amazon cutting 30,000 since October) and asks a question that almost nobody in a position of authority is asking: are these layoffs actually caused by AI, or is AI being used as narrative cover for financially motivated cuts?

He cites Deutsche Bank analysts who coined the term "AI redundancy washing" to describe companies attributing job cuts to AI capabilities that haven't actually replaced the work. The Randstad CEO told CNBC that those 50,000 tech job losses were driven by market uncertainty, not AI. Meanwhile, 89% of executives in a separate NBER survey report no measurable effect of AI on their company's labor productivity over the past three years.

McDonogh's framing is about accountability. "AI enabled it" is a mechanism description, not an accountability statement. Who made the decision? On what evidence? Were the productivity gains real, or projected? Were the displaced workers told the truth? These are the questions that matter, and almost nobody with a platform is asking them.

Where to find them: autonomy.work / "Post-Work" (Bloomsbury, 2025)

Here's a question nobody in recruiting is asking: if AI actually delivers the productivity gains everyone claims it will, where do those gains go?

The Autonomy Institute, a UK think tank co-directed by Will Stronge, has done the most serious research on one answer: reduced working hours. Their data suggests that AI-driven productivity gains could enable roughly 8.8 million UK workers to move to a four-day work week without any loss in output. The Iceland and UK trials they've studied bear this out.

Hester and Stronge's book "Post-Work" goes further. It challenges the foundational assumption that work is inherently good for you, that identity should be derived from employment, that "any job is a good job." These aren't fringe ideas. They're part of a long intellectual tradition (one that includes the concept of otium, which I've written about in earlier editions) that asks what human flourishing looks like when it's not organized around the sale of your time.

The reason this matters for the labor conversation right now: every single displaced worker is being told the answer is to retrain, upskill, and re-enter the labor market as fast as possible. That might be practical advice for individuals. But nobody is asking whether the labor market they're re-entering is structurally capable of absorbing them, or whether the entire frame of "get another job" is adequate to the moment.

I looked hard for recruiting leaders or HR thinkers who are engaging with this level of structural analysis. The honest answer is that most of what I found stays inside the pendulum model: it's a down market, it'll come back, post your content, build your brand, wait for the swing.

The closest voice I found to actually bridging the gap is Hung Lee and his Recruiting Brainfood newsletter. Hung doesn't write the structural analysis himself, but he curates toward it. He's been linking to the Citrini Ghost GDP essay, engaging with industrial policy, and pushing back on Marc Andreessen's claim that AI isn't impacting employment. His take is sharper than most in the space: AI doesn't directly replace jobs so much as it acts as a persistent suppressant on hiring demand and a pressure on existing payroll. That's closer to the truth than anything I'm seeing from the "AI won't replace recruiters" crowd. If you're in recruiting and you only subscribe to one newsletter (other than this one), Brainfood is the one. Hung is also traveling through East Asia right now doing "How to Hire" webinars in each country, which is the kind of ground-level global perspective our industry desperately needs.

But even Brainfood is still largely operating within the recruiting frame. The voices in this edition are operating outside it. That's the gap. The recruiting world doesn't yet have language for what's happening at the structural level. The entry-level pipeline is collapsing (a 16% employment decline among 22-to-25 year olds in AI-exposed roles, per Harvard Business School). Companies are recording their employees' workflows to train the systems that will replace them (Meta's Model Capability Initiative, which logs keystrokes and screen captures). 31% of Gen Z workers now report outright anger toward AI, up from 22% a year ago. The generation that was supposed to be AI-native is watching it close the doors they expected to walk through.

If you're in recruiting and you're only reading recruiting content, you're watching the pendulum. These are the people watching the wall.

I don't want to leave you in the dark. The structural picture is heavy, but there are real things happening that point toward something better. Not in the "don't worry, the market will correct" sense. In the "people are actually building alternatives" sense.

IBM is going the other direction. While every other tech giant cuts entry-level headcount to fund AI infrastructure, IBM tripled its entry-level hiring in 2026. Their argument: eliminating junior roles saves money now but destroys the pipeline that produces experienced workers and leaders five years from now. AI tools still need human direction, judgment, and context. This is a company placing a real bet that the current slash-and-burn approach is short-sighted. They might be wrong. But they're the only major tech company testing the hypothesis.

Workers are organizing, not just complaining. The Workers Lab project I mentioned earlier is small, but it represents something new: displaced tech workers forming a cooperative to build collectively rather than competing individually for shrinking seats. Twenty technologists and forty volunteers, organized in two months. It's not a movement yet. But the instinct behind it (that the answer to displacement might be collective ownership rather than individual retraining) is a seed worth watching.

Legislators are starting to act. Connecticut passed Senate Bill 5 in April, requiring employers to notify workers when AI is used in hiring, promotion, or termination decisions. It also funds AI literacy programs and creates a regulatory sandbox for testing new tools under oversight. It's one state. It's imperfect. But it fills a vacuum that the federal government has shown no interest in filling, and it treats AI displacement as a policy problem rather than an individual one. Other states are watching.

The four-day work week has real data behind it now. The Autonomy Institute's research, the Iceland trials, the UK pilots: they all show that reduced hours can absorb productivity gains without reducing output. If AI actually delivers the efficiency gains everyone claims, the question becomes whether those gains flow to shareholders or to workers in the form of time. That's a political question, not a technical one. And the fact that it's being asked seriously, backed by data, is a reason for cautious hope.

The conversation itself is changing. A year ago, the dominant narrative was "learn to prompt or die." Today, the most widely read piece of financial writing in 2026 is a speculative essay about what happens when AI productivity gains never circulate back through the real economy. The Overton window is moving. People are starting to ask structural questions rather than individual ones. "How do I keep my job?" is giving way to "What kind of economy do we want?" That shift in the question is, itself, a form of progress.

None of this is a guarantee. The structural forces are real, and they are moving fast. But the people in this edition (and the people reading this newsletter) are proof that the response isn't just fear or resignation. It's analysis. It's organizing. It's building. That matters.

404: Work Not Found is a newsletter about what happens when the thing we organized our entire civilization around starts disappearing. If this edition resonated, send it to someone who needs to see it. And if you're not subscribed yet, you probably should be.

- Tyler