A sharp disconnect has emerged between the scale of corporate experimentation with artificial intelligence and its tangible returns. Across markets, roughly 90% of firms are piloting or researching AI, yet only 10% report generating concrete value from those efforts, according to data cited in Buenos Aires and São Paulo. In Brazil, a separate survey of shared service centres found that 35% remain at the pilot stage and just 2% operate at scale. The same pattern is visible globally: the technology is being tested everywhere, but the leap from trial to production remains elusive.

The primary obstacle is not the technology itself but the absence of strategic integration. Business advisors in Nairobi argue that leadership, not software, determines success, pointing to cases where AI was deployed to automate an approval workflow only for managers to discover that the real solution was eliminating an obsolete step. In creative agencies in Buenos Aires, practitioners describe a growing fear of “outsourcing thinking,” where junior staff present AI-generated output without having developed the underlying reasoning. A Google employee writing from a personal perspective notes that her eight-year-old daughter, having watched her mother consult ChatGPT for quick answers, now defaults to the same shortcut, bypassing the struggle that builds critical thought.

Consumers, meanwhile, are already reshaping discovery and purchasing. In Brazil, 81% of consumers say they trust AI, and 64% have used generative AI in buying decisions. Globally, 37% of shoppers now begin product searches on AI platforms rather than traditional search engines, and 41% have purchased an item recommended by an AI in the past six months. This shift is forcing brands to optimise not for search-engine rankings but for how large language models synthesise third-party reputation data. Yet the same tools are being adopted by children at a pace that outstrips adult supervision: UNICEF surveys across ten countries, including Colombia, indicate that 20 million adolescents aged 12 to 17 have used AI, with one in ten turning to it for personal advice, while governance frameworks remain largely silent on child-specific protections.

The governance vacuum is drawing increasingly urgent warnings. A UN-appointed panel of 40 independent experts released a preliminary report stating that AI capabilities are “outpacing both scientific knowledge and the ability of governments to adapt,” and that current science cannot guarantee the avoidance of catastrophic harm. The panel noted that task complexity is doubling every four to seven months, and that existing safety tools rely on limited testing data disclosed by companies. In response, the UN Secretary-General called for swift government action, and a new Global Commission on AI for Good was announced, co-chaired by Rwanda’s president and the CEO of Salesforce.

The next factual milestone is the first Global Dialogue on AI Governance, scheduled for 6–7 July in Geneva, where governments will be pressed to move beyond fragmented national approaches. Whether that forum produces binding commitments or merely a shared diagnostic will shape the regulatory landscape into 2027.