How AI Could Threaten Democracy — Not Your Job

What we talked about.

In this episode, Sunny Ray sits down with Vitali Golum, managing partner at Mavak Capital, to explore how AI driven automation could destabilize societies faster than any prior technological shift. Golum, who is writing a book called The Displacement Machine, argues that unlike past industrial revolutions which unfolded over generations, AI is compressing disruption into roughly a decade, leaving little time for workers, especially those in administrative and service roles, to adapt. He draws parallels to Weimar Germany, warning that economically displaced populations become vulnerable to scapegoating and radicalization. The conversation covers unemployment projections from Goldman Sachs, the breakdown of the labor capital partnership, differing safety nets across the US, Europe, and China, and why entrepreneurship may become the default skill set for future generations. Golum also discusses his own work funding robotics and automation through Mavak Capital, reconciling that with his warnings, and explains why traditional software moats are dying while physical world advantages in manufacturing and robotics are becoming more valuable. The episode closes with reflections on generational cycles and the turbulent transition period ahead.

An investment banker warns that AI could compress a century of economic disruption into a single decade, threatening democracy itself.

The questions, and the answers.

You said any country that hits 20% unemployment goes fascist, and AI is about to make that real. Can you defend that?

Each industrial revolution gave society more time to adapt, from a century down to a single generation. AI is compressing that into about 10 years. People retrained into administrative jobs like insurance and customer support are now the first to be automated away. Goldman Sachs projects a real chance of 10 to 20 percent unemployment soon. When a middle class loses its status quickly, like in Weimar Germany, that is when extremist movements take hold.

Can you walk me through the actual mechanism, not just people lose jobs and get angry, but how does that turn into a man on a podium promising order?

Losing something feels much stronger than not gaining something, which is why revolutions come from a displaced middle class rather than a lower class denied opportunity. A leader tells people their problems are caused by immigrants or another group instead of systemic economic failure. That scapegoating lets people ignore the real problem and organize around a political movement built on blame rather than solutions.

Is the danger the unemployment itself, or the speed at which AI causes it? What worries you more?

The real problem is people won't have skills to be productive in the new economy. We're breaking the historic partnership between labor and capital, since automated factories don't need labor. But if everyone is out of work, who buys the products even after production is automated? There's a lot of emphasis on automating fast, but not enough on making sure society can actually absorb that shift.

Which country breaks first, and why?

The US has strong incentive to automate fast but weak safety nets, like a small unemployment check for 26 weeks. Countries like Denmark offer a much higher percentage of prior salary plus a place to go daily for retraining and community. China has even passed laws making it illegal to lay off workers for automation without heavy severance. The US is probably at the tip of the spear for both the benefits and the negative effects.

Help me push back on yourself. Every prior automation wave destroyed categories but created more jobs. Why is AI different?

A couple years ago people said AI would create new jobs like prompt engineer, but almost nobody talks about that role anymore. The most optimistic outcome is that entrepreneurship becomes the core skill, since you can build a billion dollar company by yourself now. Gen Z, native to computing, may adapt more easily, but millennials and Gen X in the middle of 20 to 30 year careers will have a much harder transition.

You raise capital for a living and are funding robotics and automation. Are you funding the thing you're warning about?

It's can't beat them join them. Automation is inevitable, so we want companies replacing labor with robotics to also have an answer for how they'll positively affect that labor force in the short term. We're launching a robotics venture fund because embodied AI is the fastest growing tech wave, projected at a 39 percent compounded annual growth rate reaching 76 trillion across five sectors we track.

You wrote that the moat is dead. If proprietary tech is no longer a moat in an AI world, what is, and what are you underwriting now?

Software moats are dying because anything can be replicated by someone vibe coding a product over a weekend. Real moats now are proprietary data, your own models rather than sitting on top of Claude or OpenAI, unique insight, and strong customer relationships. In robotics and physical products there's still a moat because manufacturing requires capital and infrastructure that AI-only software players don't need, though AI players now spend heavily on data centers instead.