Two and a half years after ChatGPT kicked off the generative AI panic, the US labor market hasn’t collapsed. It’s actually gotten better, according to Goldman Sachs.

The bank’s occupation-level mismatch index, which tracks how well job seekers align with available positions, has declined below its pre-pandemic level. In English: people are finding jobs that actually match their skills more efficiently than they were before a chatbot could write a sonnet about supply chain logistics.

The numbers tell a complicated story

Look, nobody’s saying AI hasn’t displaced workers. Goldman Sachs estimates AI has led to a net reduction of approximately 16,000 jobs per month over the past year. That’s real, and it stings, particularly for Generation Z workers who appear to be absorbing the brunt of the disruption.

But here’s the thing. Those losses have contributed to only a 0.1 percentage point rise in the unemployment rate. For context, that’s a rounding error in the grand scheme of labor economics. The reason the damage has been so contained is that AI isn’t just eliminating roles. It’s simultaneously creating new ones, especially in areas where it augments human labor rather than replacing it outright.

Think of it like the ATM paradox. When ATMs rolled out in the 1970s, everyone assumed bank tellers would vanish. Instead, the number of bank branches grew because ATMs made them cheaper to operate, and tellers shifted to customer service and sales roles. Something structurally similar appears to be happening now, just at a faster clip and across more industries.

Goldman Sachs projects that AI adoption could displace 6-7% of the US workforce during the transition period. That’s not trivial. But displacement and permanent unemployment are different animals entirely.

What’s actually improving

The decline in the mismatch index is the headline number, but it points to something deeper. The labor market isn’t just surviving AI, it’s reorganizing around it.

The share of US professionals who freelance increased from 34% in 2014 to 38% in 2023. That trend predates ChatGPT, but generative AI tools have accelerated it by lowering the barrier for independent work. A graphic designer who once needed a full Adobe suite and years of training can now prototype concepts with AI assistance. A freelance developer can ship code faster. The tools are making smaller operations more viable.

Goldman estimates that generative AI will raise labor productivity in the US and other developed markets by around 15% when fully adopted. That’s a staggering figure. For comparison, the internet’s productivity gains took roughly two decades to fully materialize. AI appears to be compressing that timeline considerably.

The bank also notes that AI has the potential to automate tasks accounting for 25% of all work hours in the US. Globally, that could impact 300 million jobs over the next decade. But automating tasks is not the same as automating jobs. Most roles are bundles of dozens of tasks, and AI tends to handle some while making humans more productive at the rest.

What this means for investors

Goldman’s analysis carries significant implications for anyone positioning capital around the AI trade. If the labor market is genuinely healthier, not weaker, in the wake of AI adoption, it undercuts one of the biggest bear cases against aggressive AI investment: that mass unemployment would trigger a political backlash, regulatory crackdown, or consumer spending collapse.

A 0.1 percentage point unemployment bump is not the kind of number that motivates Congress to act. That gives AI companies, and the enterprises deploying their tools, a longer runway before facing serious regulatory headwinds.

The productivity story matters even more. A 15% labor productivity boost, if Goldman’s projection holds, would be transformative for corporate margins across virtually every sector. Companies that adopt AI early would see cost advantages compound over time, while laggards fall further behind. That gap creates both opportunity and risk for portfolio construction.

The freelancing trend is worth watching too. Platforms that serve the independent workforce, from payment processors to project management tools to AI-powered creative suites, stand to benefit from a structural shift that’s already well underway.

The risk to monitor is concentration of pain. Generation Z workers bearing a disproportionate share of displacement could create social and political pressure points that don’t show up in aggregate unemployment data. A generation locked out of entry-level positions because AI handles the tasks those roles traditionally performed is a slow-burning problem. It may not crash the labor market, but it could reshape consumer behavior, housing demand, and political sentiment in ways that eventually ripple through markets.