Key Takeaways
- Forget what you think you know about AI and job destruction. As economist Alex Imas points out, experts like him have been “famously terrible at forecasting” these impacts historically.
- David Ricardo, fearing mass unemployment during the Industrial Revolution, missed the “lump-of-labor fallacy” – that automation frees up capital for new services, creating new jobs.
- The real danger isn't mass unemployment, but a "messy middle" where AI automates tasks without creating enough new wealth, during the transition, to fairly compensate displaced workers.
- This leads to a "drip scenario," as Imas describes, where workers slowly reabsorb into lower-paying, underemployed roles, creating political instability rather than an acute crisis.
The Ricardo Problem: Why Economists Miss the Mark on Automation
Most founders are bracing for AI to either create a jobs boom or a robot apocalypse. But if history is any guide, neither extreme might be the actual outcome. Economist Alex Imas, speaking on the Dwarkesh Podcast, cuts right to the chase: “My view is that the individual forecasts economists like us would make... are not necessarily very useful.” Why? Because, as Imas says, “we have been famously terrible at forecasting.”
He points to David Ricardo during the Industrial Revolution. Ricardo saw machines automating weaving and feared widespread, permanent unemployment. What he missed was the "lump-of-labor fallacy." As production became cheaper, people had more disposable income. They then spent that money on new services – education, entertainment, healthcare – which created entirely new categories of jobs. The market adjusted, albeit messily, preventing the mass unemployment Ricardo predicted.
This historical blind spot highlights a core challenge for ambitious builders today: the future of work with AI might not look like a simple replacement problem. The past suggests a dynamic, complex dance between automation, wealth creation, and new job emergence. But this time, there's a crucial difference.
The "Messy Middle" Isn't Mass Layoffs, It's Quiet Decay
The real risk with AI isn't a sudden, visible wave of mass layoffs. Instead, Dwarkesh Patel and his guests articulate a far more insidious threat: the "messy middle." This is a scenario where AI does automate jobs, potentially many of them, but crucially, it doesn't create enough new wealth during that automation process to compensate those losing out. "AI makes it possible to automate jobs such that many people are losing their jobs," Patel explains, “but it doesn't create enough wealth... to basically pay off the people who are getting laid off.”
This isn't a Pareto improvement where everyone benefits. Instead, it leads to what Alex Imas, referencing Molly's excellent essay, calls a "drip scenario." It's not about millions of people jobless; it's about millions of people slowly shifting into lower-paying work, or underemployed roles that don't match their skills. Phil Trammell adds that this happens when capital gets productive enough to automate jobs, but not so productive that interest rates spike and prices for AI-produced goods plummet, which might otherwise generate more wealth and opportunity. This slow, steady erosion of earning power and professional purpose, rather than a catastrophic event, is far more likely to spark long-term political instability and social unrest. It's a quiet decay, not a sudden collapse.
What to Do With This
Don't plan for a sudden robot takeover or an overnight jobs boom. Instead, prepare for the "drip scenario." This week, audit your talent strategy: identify the three roles in your organization most susceptible to task automation over the next 12-18 months. For these roles, brainstorm concrete ways to re-skill or augment these employees to higher-value, more creative, or interpersonal functions that AI can't touch, rather than just replacing their automated tasks. Your goal isn't to cut costs via AI automation, but to proactively elevate your human capital and avoid creating a mini "drip" within your own walls.