A question came across my feed this week that I couldn't stop thinking about: As AI becomes more advanced, will humans have to revert to socialism because there simply won't be enough jobs? Most people would dismiss that as alarmist. I think it's one of the most important questions of the next two decades; and we're not taking it seriously enough.
Let me be clear about where I'm coming from. I'm a CISO and a technologist. I've spent 25 years building security programs inside financial institutions. I'm not a Luddite. I believe deeply in the power of technology to create value and improve lives. And it's precisely because I understand what AI can actually do; not the hype, the actual capability; that this question keeps me up at night.
The jobs conversation about AI is different from every prior automation wave. And if we don't understand why it's different, we'll keep applying the wrong mental models to the problem.
Why This Wave Is Different
Every prior automation wave; the loom, the assembly line, the computer; displaced workers in specific, bounded categories. Agricultural machinery reduced farm labor. Factory automation reduced manufacturing jobs. Accounting software reduced bookkeeping roles. But each time, a new category of knowledge work absorbed the displaced workers. Farmers became factory workers. Factory workers became office workers. That transition was painful and uneven, but the pattern held.
AI breaks that pattern. For the first time, we are automating the cognitive layer; the layer that historically absorbed displacement from below. Legal analysis, medical diagnosis, financial advisory, content creation, software development, customer service, security analysis. These are not narrow capabilities. These are the job categories that educated workers have retrained into for fifty years.
"If AI can do the cognitive work that displaced workers have always retrained into, where do those workers go? That's not a rhetorical question. It's an economic emergency in slow motion."
MIT economist Daron Acemoglu has argued that AI could hollow out the middle class faster than new job categories can emerge. Stanford's Erik Brynjolfsson counters that productivity gains will ultimately create demand for new work we can't currently imagine; as has historically been the case. Both are serious economists with serious arguments. The honest answer is that nobody knows which scenario plays out. What's clear is that the magnitude of risk justifies the urgency of the conversation.
The Wealth Concentration Problem
Here's what makes the AI displacement scenario uniquely dangerous compared to prior automation waves: the productivity gains accrue almost entirely to capital, not labor.
When a factory installs a robot, the owners of the factory capture the efficiency gain. The displaced worker does not. Historically, that imbalance has been partially offset by workers finding new jobs and contributing to a broader consumer economy that funded the next wave of growth. That feedback loop depends on labor having somewhere to go.
If AI systematically eliminates the destinations; if both the factory floor and the knowledge work office are automated; the productivity gains pool at the top of the capital stack. The owners of the AI systems capture the value that was previously distributed across millions of workers. That's not a political statement. It's arithmetic.
AI creates enormous wealth. The question is whether that wealth is distributed broadly enough to sustain the consumer economy that makes it valuable in the first place. Concentrated wealth in a low-employment economy is not stable. Henry Ford understood this; he paid his workers enough to buy the cars they built. The AI economy hasn't figured out its equivalent.
Is Socialism the Answer? The Policy Landscape
The question about socialism is really asking: what redistribution mechanisms can sustain a society where traditional employment no longer provides the economic foundation for most people? And the honest answer is that "socialism" as a monolithic concept is the wrong frame. What's actually being debated is the degree and mechanism of redistribution; a spectrum, not a binary.
Here are the serious policy responses being explored:
Universal Basic Income
A guaranteed income floor for every citizen regardless of employment status. Piloted in Finland, Stockton CA, Kenya, and elsewhere. Andrew Yang campaigned on it. Sam Altman at OpenAI has personally funded UBI research. Results from pilots are generally positive on wellbeing metrics but scale questions remain unanswered.
AI Productivity Taxes
Taxing the productivity gains of AI systems to fund displaced worker programs. Bill Gates has proposed "robot taxes." The political economy is difficult; capital is mobile and will resist; but the structural logic is sound. If AI captures value previously held by labor, redistribution requires capturing some of that value.
Shortened Work Weeks
Distributing available work across more people with fewer hours each. France's 35-hour week and various four-day week pilots show productivity can be maintained. This is more politically viable than UBI but doesn't address scenarios where total available work falls below the threshold to sustain full employment even with redistribution.
Public Employment Expansion
Government as employer of last resort; funding work in care, education, infrastructure, and community services that the market underprovides. More aligned with traditional social democratic models than full socialism. The challenge is funding at scale in an era of strained government budgets.
None of these solutions are "socialism" in the classical sense of state ownership of production. They are all market economies with varying degrees of redistribution layered on top. The question isn't whether the West becomes Venezuela. The question is whether current redistribution mechanisms; designed for an era of full employment; are adequate for an era of structural underemployment.
The Timeline Is Everything
Here's the variable that makes this so hard to plan for: we don't know the speed of displacement. The policy response to gradual disruption over fifty years looks very different from the response to rapid disruption over a decade. Societies can adapt to slow change. They struggle with fast change. And AI capability is not moving slowly.
Augmentation Phase
AI primarily augments skilled workers. Productivity gains are real but employment remains stable. The risk is invisible; displaced jobs appear as slower hiring, not mass layoffs.
Substitution Phase
Agentic AI systems handle complete job functions, not just tasks. Junior and mid-level knowledge work roles are most exposed. Wage pressure intensifies for remaining roles as labor supply exceeds demand in cognitive work categories.
Structural Reckoning
If economic policy hasn't adapted by this point, political instability becomes the forcing function for change. History suggests societies don't restructure smoothly under that kind of pressure.
What Leaders Need to Do Now
I'm a security leader, not a policy maker. But I spend my career thinking about systemic risk; and this is a systemic risk that is moving faster than institutions can respond. So here's what I think needs to happen:
Corporate leaders need to own the transition, not just the productivity gain. If your AI implementation eliminates a job function, what is your obligation to the people in those roles? Reskilling programs, extended transitions, income support; these aren't charity. They're the cost of sustainable growth in a society that depends on broad-based consumption to buy your products.
Governments need to start the policy conversation now, not when the crisis arrives. Every major economy needs a serious working group on AI employment impact; not to slow AI, but to design redistribution mechanisms before they're needed in emergency mode. The window for thoughtful policy is closing faster than most legislators realize.
The technology community needs to stop dismissing the concern. The "new jobs will appear" argument is historically grounded but not guaranteed. Saying "it worked before" isn't a policy. The scale and speed of current AI capability development is genuinely different from prior automation waves, and the intellectual honesty to acknowledge that uncertainty is the minimum standard for serious participation in this conversation.
"The goal is not to stop AI. The goal is to make sure the extraordinary wealth it creates doesn't concentrate in so few hands that the social contract that makes civilization possible breaks down."
The Real Question
I don't know if societies will "revert to socialism" as AI advances. I don't think that's the right frame. What I know is that the current economic model; where employment is the primary mechanism for distributing the gains of productivity; was designed for a world where human labor was the scarcest input. We are building a world where intelligence itself becomes abundant and cheap. That changes everything downstream.
The societies that navigate this well will be the ones that start designing for it now; building redistribution mechanisms, education systems, and social contracts suited to an AI-abundant economy; rather than the ones that wait until the pressure becomes impossible to ignore.
That's not socialism. That's governance. And it's past time we started treating it with the urgency it deserves.
This post reflects personal views. All figures cited from publicly available research and policy literature.