There are many signs of an AI jobs crisis:
-
Entry-level job postings have dropped sharply over the past four years
-
Cross-industry and large scale layoffs are being blamed on AI
But unemployment remains low, many “AI layoffs” turn out to be ordinary cost-cutting with a new label, and a majority of senior executives say they expect AI to increase entry-level hiring.
Both insights in this narrative can be true. The better question is what the current calm is actually telling us. My read is that it tells us much less than most people assume.
Today’s data is a poor predictor of tomorrow’s disruption. General-purpose technologies move slowly, then all at once. Current effects reflect early adoption, not the technology’s eventual reach. Plausible scenarios range from modest gains with little job loss to displacement larger than the China trade shock. Most workers would find new roles eventually, but the transition would be painful.
-
Clumsy execution, not limited capability, is the real brake. About 40 percent of work tasks can already be automated or augmented, yet only a third of AI pilots succeed. The usual reasons are poor data and untrained employees. Leaders who read weak results as a verdict on AI will be caught flat-footed as cheaper, purpose-built agents close the gap.
-
Replacing human capital is a strategic trap. AI can gather expertise that used to be scattered across people, but cutting investment in those people trades short-term savings for long-term fragility. Tacit knowledge never appears in a job description. Training is how a firm builds and protects the knowledge that sets it apart.
-
Redesign of organizations is underway. There is no senior talent without junior talent. Leading firms are rebuilding first jobs to pair core work with structured exposure to sales, product, and customers. The result is a pipeline of future leaders with experience their current managers never had.
-
Corporate leaders are skeptical of effective disruption based on technology alone. Many have been in this rodeo. Companies will choose the lowest-cost option, and sometimes that will be AI. Managing the broader disruption will require proven value, more people investment including retraining, and changes in the safety net.
For boards and executives, the right posture is the same one we take toward any low-probability, high-impact risk. Prepare now, not because disruption is certain, but because being surprised by it is far more expensive.
Conclusion
There are many questions worth asking now. For example, five years from today, where will most of the economic value created by AI reside—in models, infrastructure, applications, proprietary data, or entirely new business models? Are today’s AI pilots failing because of the technology, or because organizations are applying AI to processes designed for another era? Where does our tacit knowledge live, and what happens when we automate the roles that accumulate it? And what should entry-level jobs look like if we’re building for 2036 rather than 2026?
These aren’t simply technology questions. They are questions about value creation, organizational design, human capital, and competitive advantage.
The companies that benefit most from AI may not be those that adopt it fastest, but those that understand what to automate, what should remain human, and where AI fundamentally changes the economics of their business.
AI is moving quickly. The harder challenge is deciding what we want our organizations to look like when it gets there.