The first measurable cracks are appearing in the labour market for young graduates. In the United States, the unemployment rate for recent college graduates has historically tracked just above the rate for all graduates; that gap is now widening, and analysts point to artificial intelligence as a primary cause. China, where youth graduate unemployment has been rising for years, has seen the trend accelerate. Neither economy is shedding jobs overall—headline unemployment remains near multi-decade lows on both sides of the Pacific—but the divergence at the point of entry is sharp enough to draw the attention of central bankers and labour ministries.

The mechanism is not mass layoffs but a quiet thinning of demand for the cognitive tasks that once constituted a first professional rung. Large language models now draft press releases, write code, generate advertising copy, and produce photorealistic images, compressing workflows that previously required junior hires. The creative industries offer a vivid stress test: three of five regional winners of the 2026 Commonwealth Short Story Prize stand accused of using AI, and newsrooms from Delhi to Bogotá report that AI-generated content is crowding out entry-level photojournalism and copywriting. The boundary between assisting human decisions and making them has shifted, as a pianist and educator in India observed, and the distinction between human and machine output is no longer reliably detectable.

Yet the same technology is elevating a different human faculty. At the Cannes Lions festival, marketing chiefs from Autodesk, Kimberly-Clark, and Zoom described a pivot from accelerating AI adoption to cultivating “taste”—the discernment to know when not to use the tool. As content production becomes democratised, they argued, competitive advantage migrates to the human insight that drives better outcomes. OpenAI’s chief economist, speaking at the European Central Bank’s annual retreat in Sintra, reinforced the point: exposure of a task to AI does not mean substitution, and the personal computer ultimately made his economist father more productive, not redundant. In Latin America, the same logic is fuelling a digital-entrepreneurship boom; Colombia is emerging as a net exporter of Spanish-language knowledge products, with the market for digital courses and content projected to reach USD 98 billion by 2030.

Governments are watching the same data through a strategic lens. The EU has operationalised its AI Act, setting risk-based rules that may become a global compliance benchmark, while Washington has tightened export controls on advanced chips and, in one contested case, ordered a leading AI firm to suspend foreign access to its most powerful models. For India, which possesses vast digital infrastructure but remains a consumer rather than a builder of foundational AI, dependence on external ecosystems is increasingly framed as a sovereignty risk. The next factual milestone is the release of quarterly labour-force surveys from the US Bureau of Labor Statistics and China’s National Bureau of Statistics, which will show whether the graduate unemployment signal strengthens into a structural trend or fades as a cyclical adjustment.